Due to concerns regarding power quality and the increasing electric power system complexity, Static Synchronous Compensators (STATCOMs) are widespread in power systems for reactive power compensation and voltage support. For medium- and high-voltage applications, the multilevel converter is the most common option for STATCOM systems. Additionally, the Delta-connected Cascaded H-bridge (Delta-CHB) is a widely adopted multilevel converter topology for STATCOM systems. This work delves into the Delta-CHB STATCOM control strategy, proposing dynamic modeling for this converter and providing a methodology for control tuning for the several controllers present in this topology. This work also explores the normalization strategy and its effects on circulating current. Moreover, the study addresses the second-harmonic ripple in converter capacitor voltage and describes its relation to the third-harmonic circulating current, inherent in the converter operation through mathematical expressions. Validations are conducted through both simulations and a prototype with three submodules per phase. The presented ripple generation mechanism model exhibits differences smaller than 4.5% compared to the simulation. Finally, the presented control tuning methodology was tested through simulations in two STATCOM systems at different power levels, and in a prototype, validating the control tuning strategy.

This work proposes a method of analysis of printed Magnetic Coupled Wireless Power Transfer (MCWPT) systems operating in a few MHz based on Circuit Model (CM). Initially, the effect of resonant surfaces on wireless energy transmission and the advantages of their use are discussed. Therefore, the resulting increase in efficiency is discussed based on the Magneto-Inductive Waves (MIW) theory. Furthermore, the feasibility of using metasurfaces in Wireless Power Transfer (WPT) systems where the transmitter and receiver are misaligned or are coplanar is analyzed. Finally, possibilities for optimizing the impedance of WPT systems to increase transmission efficiency are also presented.

An analytical model is proposed to calculate the self-inductance of printed inductors, their characteristic resistance, and the mutual inductance between coils. The presented formulation for calculating the mutual inductance between inductors is based on Neumann's formula. Due to the generality of this approach, it can also be applied to more complex structures, such as metasurfaces. In the analysis of the MCWPT CM, it is considered that the unknown current along the microstrip is considered as a single value distributed in each modeled coil. First, computational aspects related to model implementation are discussed. Then, the calculated results are compared step by step with commercial software based on electromagnetic methods in the frequency domain. Then, the microstrip model is experimentally validated with measurements of transmission coefficients from an MCWPT printed in FR4, operating at 24 MHz. A new design based on printed square coils is also proposed, but now with reduced size and working at 13.56 MHz. A frequency domain analysis on how the frequency and receiver position on the metasurface affect efficiency is done. Then, the variation of the input impedance and the current distribution on the surface of the metamaterial are analyzed in relation to the variations in the position of the receiver and in the charge rate on the resonant surface. For this, the MIW theory is used in a specific application: Passive Position Tracking in Dynamic MCWPT Systems. Finally, it discusses the importance of knowing the current distribution and impedance optimization parameters for an MCWPT system.

The results show that the proposed method is valid for modeling printed inductors. In addition, the model proved to be computationally more efficient for the analysis of large systems when compared to commercial full-wave simulation software, and the practical results point to conformity.

The present work deals with the application of the Monte Carlo method as a tool for planning and controlling the maintenance of sirens in tailings dams in mining. Strategies that impose constant improvement in maintenance planning and execution, especially in the issue of dam sirens, are fundamental, given the lethal potential of an accident within mining companies. Thus, in this work, the application of the Monte Carlo Method was investigated to analyze the predictability of maintenance on siren poles in mining tailings dams. Specifically, a literature review was carried out on mining processes, tailings dams and maintenance control and management processes; the scope of application of the Monte Carlo method in industry was investigated considering its application for maintenance planning and control; and the method was applied to estimate the resources dedicated to the maintenance of siren posts in mining tailings dams, considering the most critical maintenance scenario.Through a quantitative and qualitative approach focused on describing, interpreting and understanding the perception of safety management in mining tailings containment dams, a bibliographic survey was carried out, seeking to meet a demand for information on the maintenance management of dams in the mining industrial branch from the national scenario that was based mainly on the comparison of the classification criteria of dams in Brazil and the maintenance perspectives considering the legislation and measures adopted to promote safety in dams. The case study was based on data from tailings containment dams of a mining company operating in the state of Minas Gerais as a real source for the application of the Monte Carlo Method.The main result found was the confirmation of the effectiveness of the Monte Carlo Method for the analysis of maintenance predictability in siren poles in tailings dams. Efficiency was verified when tracing the trend of scenarios of future failures, based on the history, with statistically predicted oscillations through the generation of random numbers. It was possible to conclude that the methodology proposed and applied to the set of failures of siren posts in mining tailings dams with the aim of predicting failures, for pessimistic and optimistic scenarios, was validated and allows directing maintenance in an assertive way. And it also allowed us to understand that the increase in the number of critical failures after carrying out the practical emergency simulations, may come to characterize a technically inadequate design specification for the applicability condition of the electronic components of the sirens.

Monitoring the energy consumption of induction motors is an important task in industry to ensure cost-effective operation. The motor efficiency is conveniently obtained from the electromagnetic torque and the rotation speed. Torque is easily calculated from stator voltages and currents and depends only on one machine parameter, the stator winding resistance, Rs. To accurately estimate the electromagnetic torque, Rs is usually estimated online. Among the various Rs estimation methods available in the literature, the DC injection method is one of the most successful. It has a high accuracy and an acceptable level of intrusion. However, the method causes oscillations of electromagnetic torque and speed in the machine that are highly undesirable for several industrial processes. It is well known that the zero-sequence components of the stator current do not contribute to the electromagnetic torque and do not cause oscillations in the electromagnetic torque. Being calculated from a simple first-order model, the zero-sequence stator current can be used to estimate Rs. The central proposal of this work is to use zero sequence components of stator voltage and current both in Machines with Open-End Winding Induction Machine, OEW, and in the Induction Machine directly connected to the network to estimate Rs. In the OEW configuration with a single bus and under certain conditions, zero-sequence currents flow. These currents are used here for the estimation of Rs. In the configuration with the machine connected to the mains, the central terminal of the machine is connected to the neutral terminal of the source, to promote the circulation of zero sequence current commonly existing in the mains. Another strategy adopted here is the implementation of a pulse generator circuit to synthesize this component without depending on the existence of zero-sequence voltage in the mains. Among the zero-sequence injection proposals, pulse injection had a more satisfactory performance, with the estimation of Rs with a practically zero error. Through the experiments, the impact of using measuring sensors with a scale much larger than the voltage and the impact of the sampling rate on the estimation were observed.

In heavy industries such as steel and mining, for example, frequency inverters must be able to operate in the medium voltage and high power range. Thus, multilevel inverters have a significant importance in drives in this kind of operation. Its advantages include a better frequency spectrum of output voltages and currents, reduction of switching losses and the stress suffered by electronic switches. In this sense, the Neutral Point Clamped (NPC) converter fulfills this role well. This topology is consolidated in the market, and is very important in high power drives. It has some variations, such as the NPC Type G, which provides five levels of phase voltage at the output of the converter. The control and switching techniques applied to these inverters are a determining factor in the behavior of the synthesized voltage. This work proposes the development, implementation and experimental evaluation of a switching strategy that uses the current hysteresis. The drive is performed digitally, wich means that, the signals measured and employed in the control are sampled. Thus, it is common to use the sampling frequency to limit the switching frequency of electronic switches, since the maximum actuation frequency is half the sampling frequency. Low sampling rates, on the other hand, decrease control effectiveness, and cause problems in current hysteresis behavior operating in multilevel converters. In this sense, this research proposes a way to implement the current hysteresis with high sampling rates, using a frequency limiting circuit in the control signals of the electronic switches. This technique dissociates the sampling frequency from the actuation frequency, improving the operation of the modulator and reducing the total harmonic distortion of the inverter voltage and current. These characteristics are confirmed by results collected in the implementation of the drive through computer simulation and laboratory experimentation.

Nowadays, the interest of the scientific community in wireless energy transmission through rectennas has shown a significant increase due to the development of new technologies, including new energy demands for devices used in the internet of things. However, the safe power limits of the RF source and the low energy density as it moves away from the available source limits the application of rectennas. The sensitivity of the rectenna for a given power density can be improved through the adoption of several technical strategies, among which the increase in gain and the optimization of the radiation and total efficiency of the antenna stand out. The contribution of this work is the development of a rectenna system, for feeding a wireless sensor, with printed antennas of high gain and high total efficiency. FR-4 and RT/Duroid 5880 substrates are used in different geometries in the construction of antennas and rectifiers in order to obtain the best performance. CST® is used to simulate and optimize the antenna parameters, and ADS® is used to develop the rectifiers. The voltage doubler topology is adopted for the rectifier in order to obtain voltage levels appropriate to the load’s needs. The designed samples are optimized to obtain S11 values below -10dB at the frequencies of interest. The constructed antennas and rectifiers presented results close to those obtained in the simulations. The rectenna system was validated by measuring the current and voltage delivered to the load with the rectenna at varying distances from the transmitting antenna connected to the RF source. There were small variations in the rectenna output for loads with impedance greater than 1 kΩ, but for loads with impedances between 10 Ω and 240 Ω the voltage levels were reduced to very low values, compromising both the rectenna performance regarding the proper functioning of the load. Therefore, a buffer stage was developed at the rectenna’s DC output to improve the ability to supply loads with lower impedances. The conversion efficiency of the designed rectenna at a load of 820 Ω was measured and is compatible with other results available in the literature. Furthermore, experiments carried out with boost converters show an improvement in the direct current power supply capacity for the proposed wireless sensor module.

The high demand for portable devices, electric vehicles, and non-dispatchable power generation sources such as solar and wind drive the search for more efficient methods of storing electrical energy. Among the ways to store electricity, using batteries becomes one of the best options, due to the greater ease of designing and applying this technology. Several chemicals can be used in batteries, but lithium-ion technology has been highlighted by greater efficiency and service life. Increasing the life of a battery is one of the main current themes, since in applications such as electric vehicles, the battery corresponds to a large part of the initial investment and is usually replaced after losing about 20% of its nominal capacity. To use electric vehicle batteries for longer, and contribute.To use the batteries of electric vehicles for longer, and contribute to increasing the life of the batteries, techniques that use the technology of intelligent batteries arise. Intelligent battery refers to distributed methods of managing a battery, reducing the impact of a centralized topology, in which individual management is more complex. This work aims to study techniques of balance of state of charge (SOC) and state of health (SOH) in smart batteries. To achieve this goal it became necessary to develop a model of a battery with representation of its degradation in PLECS software. In addition, an average model of the dc/dc boost converter was implemented to reduce the complexity of the model and allow long-term simulations. A SOC balancing technique, with fixed proportional gain, found in the literature, is implemented. Soon after, the technique found in the literature has its fixed gain changed to a variable gain, which will depend on the current data of the battery. A cascading SOH balancing technique is presented and its control action is applied as an offset in the SOC balancing technique. A case study is conducted on ten batteries, with different capacities and charge states. The SOC balancing technique with fixed gain presents good results, but in the balancing technique with variable gain balancing occurs faster and ensures a smaller permanent error. SOC balancing is a dynamically faster-balancing technique than SOH balancing, but the use of SOH balancing techniques can postpone the replacement of a battery.

The continuous increase in power demand require higher capacity for electric power transmission. Among the alternatives, the installation of series compensation on long High Voltage transmission line has the advantage of increasing the power transmission capacity, improving the stability of the system, reducing the need for voltage control equipment, in addition to being more economical than another options for increasing power transmission. However, some problems might happen in the system when using this alternative, mainly related to protection. The distance relay is one of the main equipment for electrical systems protection, it is widely used in the main protection in longer transmission lines and in the rear protection in the shorter ones. The philosophy of this protection is based on the measurement and evaluation of the impedance of the short-circuited section, which is proportional to the distance to the fault point. However, line series compensation and capacitor bank protection change the relay trigger parameters. This work aims to improve the accuracy in identifying the type of fault and determining the zone of action of the distance relay in compensated transmission lines using artificial neural networks. For this, a batch of fault cases was simulated in lines of the North-South trunk of the Brazilian electrical system, with series capacitors at both ends. The angle of incidence, the fault resistance and the distance of occurrence were varied to generate the ten types of short circuit. The results allow to compare the performance of neural networks in different scenarios and between the conventional relay and the proposed model based on neural network.

This work presents a numerical approach for computing and optimizing the electromagnetic fields generated by overhead power lines, by means of the Method of Moments and the Ellipsoidal Algorithm. The Method of Moments is based on an integral formulation and allows the unbounded problem to be entirely defined at its boundary, which reduces the number of variables in comparison with other methods. The proposed methodology focuses on a dual perspective for calculating both the electric and magnetic fields, and can be used to find the potential gradients not only at the surface of the conductors, but also at any point of interest in space. The work also proposes the analysis of the influence of grounding wires, and of different interpolation functions and boundary element quantities in the resultant field distributions. Lastly, the method of moments is coupled with the ellipsoidal method of optimization, with the goal of minimizing the electromagnetic field intensities at ground level. Multiple design and security constraints related to the original project of the line are considered in the optimization process, and the new bundle positions incur in reduced electric and magnetic field profiles at ground level, without surpassing the imposed critical field limits on the surface. The obtained results suggest the presented methodology is robust and can be applied on the design and optimization of overhead power lines.

This work presents a study for the reduction of losses in electrical power systems. In view of this objective and considering the increasing insertion of distributed generation, the problem of optimal allocation of these sources is considered. Given a specified number of generations distributed with fixed capacity, it is desired to define in which buses they should be positioned in order to minimize system losses. This is a binary integer optimization problem, where the variables can only take on two different values: 0 does not install to the bar associated with the variable, or 1 does. For the case studies, there are two subtransmission systems: IEEE14 buses and 30 buses considered, in addition to the allocation of distributed generations with a capacity of 1 MW. In the IEEE 30 bus system, considering the allocation in twenty-nine of the thirty buses, there are more than five hundred million possible combinations. Combinatorial optimization problems, with a very high number of solutions, are difficult to solve. The computer program Matpower was used to enable the calculations and analysis of the power flow. From the results obtained, with the active power losses of the transmission lines, computational methods of binary integer optimization were evaluated for the best possible allocation of the distributed generation, in order to minimize these losses. Binary integer nonlinear optimization methods, based on particle swarm algorithms and on the differential evolution algorithm, were compared to determine the most adequate method to deal with the problem. As a result, an optimized reduction of the active power losses of the systems was obtained, for values that depend on the limit of sources of 1 MW, with low computational time, showing the feasibility of the presented methodology. In addition, the results obtained from the analyzed systems were compared with results found in the specialized literature.

The disturbance observer based control is a control tool applied in several industrial
sectors, with the purpose of rejecting disturbances not measured in the process. Since
its conception, several implementations have been shown to be efficient, both for systems
represented in the time domain as well in the frequency domain. This work presents
a study on this technique as well as its application. The application of the disturbance
observer based control for systems with one input and one output, of minimum
and non-minimum phase, consists of a well-established technique. For systems with
multiple inputs and multiple outputs with time delay, there are two challenges that
must be analyzed carefully. The first challenge is to calculate the inverse of the nominal
model of the plant. The second challenge is the ideal synthesis of the disturbance
observer filter to achieve a tradeoff between the rejection of disturbances and the attenuation
of measurement noises. The proposal presented here is a strategy that uses
equivalent transfer function to address the challenge of calculating the inverse of the
nominal model of the plant that is compared with other strategies found in the literature.
The equivalent transfer function approximates the inverse of a multivariable
model by a matrix whose elements are the inverse of first order plus time delay transfer
functions, which are simpler to deal with. In order to treat the ideal synthesis of the
disturbance observer filter, a formulation based on a multi-objective optimization problem
is proposed. Through a multi-objective evolutionary optimization algorithm, it
is possible to obtain a set of efficient solutions with different tradeoffs between the two
objectives. This same synthesis formulation was applied for the problem of systems
with one input and one output in the frequency domain and to treat multivariable
systems in the time domain. The technique is initially applied to a minimum phase
problem represented in the frequency domain together with the synthesis of the PI
controller. Subsequently, two examples of multiple input and multiple output systems
with time delays represented in the frequency domain are treated. Finally, the technique
is applied for two time domain systems. The results found show that the proposed
strategies for synthesis of disturbance observer based control are effective. In the case
of multivariable systems in the frequency domain, the proposed use of the equivalent
transfer function, both for synthesis and for implementation, proved to be efficient,
presenting better results than other existing techniques in the literature.

This work presents a novel concept of an optimized digital hysteresis current control for three-phase three-level active rectifiers that are aimed to high power industry applications. The proposed switching strategy is based on a digital implementation of a hysteresis current control working with low sampling rates. The reduced sampling frequency intentionally limits the converter switching frequency and semiconductor stresses decreasing the risks of converter semiconductor failures, without deteriorations of the quality of voltage and current. It also reduces the spectral harmonic spreading of a classical hysteresis current control. The control system is conceived to limit the current ripple, stresses and losses in the power switches by an adequate choice of the sampling rate and an appropriate design of the input inductors. It leads to an optimized size, weight and costs of the input inductors. Furthermore, the proposed control is applied to a prototype of a high power and low cost three-level rectifier, which consists of two active semiconductor switches and four power diodes per phase. The converter shows good quality of voltage and current waveforms and along with its control system presents lower maintenance and acquisition costs in mission critical applications. Simulation and experiments support and confirm the proposed ideas.

Hybrid semiconductor modules consist of solutions that connect in parallel silicon IGBT with high speed switches of wide-bandgap (WBG) materials, providing high performance converters and better cost benefit ratio. These modules are presented as an alternative to increase efficiency in relation to silicon technologies, but with an aggregate total cost lower than the full WBG modules. Since it is a recent technology, the development of hybrid modules composed of Si-IGBT and SiC-MOSFET or GaN-HEMT is still restricted to research laboratories, which have made progress on the main challenges for making the technology feasible. Among the research topics, it is possible to highlight the requeriment of behavioral modeling for traditional power electronics simulation environments. Therefore, this work proposes a strategy to model a parallel connection of switches and diodes that, through look-up tables extracted from datasheets, allows a coupling between the electrical and thermal domains of the semiconductor devices. Based on this modeling, this work also proposes a device selection methodology to compose a hybrid pair formed by a Si-IGBT and a SiC-MOSFET, which are used to drive a 440 V and 300 HP induction motor. Among the solutions analyzed, two hybrid solutions with potential for loss reduction were chosen and compared to pure silicon and SiC solutions of the same rated current available on the market. The comparisons are made from an actual daily mission profile for the electric drive of an exhaust fan. The results showed a potential reduction in daily energy losses of 45 % in relation to the original silicon modules, in addition to presenting an area reduction for the SiC chip, which leads to lower cost solutions. Therefore, the selected hybrid pairs were obtained as the most cost-effective solutions considering the mission profile.

The current power systems scenario is characterized by complex structures, presence of nonlinear loads and high penetration of renewable energy such as solar and wind power plants. Then, the Static synchronous Compensator (STATCOM) emerges to perform voltage regulation, reactive power control and frequency regulation in these power systems. Since STATCOMs are hot standby devices, the converter efficiency is very important. In this scenario, modular multilevel cascade converter (MMCC) topologies have been used for STATCOM realization due to their high efficiency. To further improve the STATCOM efficiency, different strategies have been proposed in literature. However, most of the strategies require modification in the hardware. In addition, most proposals that do not require modification to the hardware have the potential only to reduce switching losses. Nevertheless, since MMCC usually employs switching frequencies in the range of 100 - 200 Hz, the conduction losses dominate. Indeed, few strategies to reduce conduction losses have been proposed in the literature. This Master Thesis aims to fill this void. This work proposes a minimum cell operation control to reduce power losses in STATCOM based on MMCC. The main principle is to bypass cells from the MMCC, according to the reactive power reference. For such, analytical expressions for the minimum dc-link voltage to keep the converter operation in the modulator linear region are derived. These expressions are used to compute the number of cells which can be bypassed for given operation conditions. Moreover, the potential of the proposed strategy for power losses reduction and the limitations of the proposed approach are investigated. The dynamic performance of the proposed scheme is evaluated based on simulations in PLECS of a 17 MVA, 13.8 kV STATCOM. Finally, an one-year energy saving study was conducted for a real reactive power profile, which presented a 7.37 % reduction in the total energy losses. The results indicated that this methodology is a breakthrough solution to reduce power losses and operational costs. The proposed technique can be applied in MMCC-based STATCOMs with more than 10 cells per arm and does not require additional hardware.

The operational analysis of the retraining Transmission Lines (TLs) makes it possible to measure their qualitative and technical aspects. This work dedicates to evaluate the lines rebuilt by HSIL - High Surge Impedance Loading technique in a permanent regime operation. The HSIL retraining techniques seek to change the electrical parameters of the lines, such as resistance, capacitance and inductance. These conditions are obtained by optimizing the position of the phase conductors. For this, three optimized overhead transmission lines, with different voltage levels and geometric configuration of conductors, are electromagnetically modeled and compared to their real and original configurations. The modeling process starts from the calculation of the electrical parameters of the TLs, such as impedance, capacitance and inductance, based on their physical and geometric characteristics. With the electrical parameters of the TLs, it is possible to obtain the surge impedance of these lines and simulate different operation modes in steady state. The empty, short-circuit and loaded operating conditions of the TLs retraining are simulated and compared to the TLs original. When empty, the Ferranti effect is attenuated at all voltage levels. In the other operating modes, the short-circuit current and the percentage voltage drop at the receiver are reduced. The performance indexes and the qualitative parameters of the lines, such as the reflection coefficients, the transmission efficiency, the reactive consumption and the voltage regulation are calculated to validate the performance evaluation of the HSIL technique. In general, the HSIL lines show operating gains on a permanent regime with consequent economic advantages.

Linear matrix inequality formulations are an important tool to tackle analysis and synthesis of control systems, especially for uncertain systems represented by polytopic models. Linear matrix inequalities are successfully applied to robust state-feedback control synthesis. On the other hand, robust static or dynamic output-feedback control synthesis are more complicated to formulate as a convex optimization problem. Different strategies for obtainingLinear Matrix Inequalities formulations for static output-feedback control synthesis requires the calculation of specific parameters, scalar and/or matrix, which directly affect the performance of the resulting controller or even the existence of a feasible solution. In this work, optimization problems are proposed to determine these parameters, aiming to obtain the optimal robust static output-feedback control system for continuous-time time-invariant linear systems. An evolutionary optimization algorithm is selected to solve the proposed problem. It is presented case studies to demonstrate the advantage of combining optimization with LMI formulations.

This work presents a study of grounding systems subjected to low frequency phenomena. The electromagnetic model of the grounding system was developed from Maxwell's equations for a configuration composed of a vertical rod and a counterweight cable. A mathematical modeling based on the Laplace equation was developed. This model was numerically evaluated using the Galerkin Free Elements meshless method, which was computationally implemented in MATLAB environment. The numerical solution was developed and applied to grounding systems in three dimensions. The results of electrical potential at ground level and resistance were evaluated for some configurations of grounding systems at low frequencies of a vertical rod buried in a homogeneous and heterogeneous soil. Parametric analyzes were also performed to properly adjust the control parameters of the numerical method under investigation. The results obtained were compared with those generated from the Method of Moments. Finally, a simulation of a counterweight cable inserted in a homogeneous soil is implemented. Results were compared with an analytical approach.

This paper investigates the problem of local stabilization for discrete time systems with time varying parameters, LPV (Linear Parameter Varying), and subject to saturating actuators. The saturation function is represented as a dead zone nonlinearity, implying the application of the generalized sector condition. Due to the limitation in the control signal it is necessary to characterize an estimate region of attraction in which any trajectory of states beginning within this region converge to the origin. Thus, convex conditions are proposed to design state feedback controllers that locally stabilize the closed loop system for a set of allowable initial conditions. The first conditions presented are based on Homogeneous Polynomially Parameters-Dependent (HPPD) Lyapunov functions. Thanks to a set of levels defined from an Lyapunov function, less conservative estimates of the region of attraction are obtained. Controller gains can assume rational structures in time-varying parameters, producing better estimates of the region of attraction as well as a broad set of stabilizable systems. The others conditions presented in this paper are based on polyquadratic functions, which is a particular case of homogeneous polynomial functions. Numerical examples are presented to demonstrate the effectiveness of the proposed methods. The first condition based on polyquadratic functions which allows to take into account the parametric dependency in the system model input matrix. The second condition guarantees a minimum contractility rate of the Lyapunov function through the adopted performance index which is λ-contractivity. A proportional integral structure is employed to ensure null tracking error for piecewise constant reference signals. Therefore, this proposal fits the design requirements of real LPV and quasi-LPV systems under saturating actuators. Experimental trials conducted on a second order nonlinear level control illustrate the potential of the proposed approach. In addition, tests indicate how the contractility rate affects the size of the estimated region of attraction. 

Modular multilevel converters have become popular in the industry in several applications such as flexible AC transmission systems, static synchronous compensators, motor control, among others. Despite of the advantages of a modular and scalable design, this class of converters are vulnerable to sub-modules faults, which can cause loss to machinery and performance. Due to this, the present work approaches the modeling and control of a modular multilevel converter using Half Bridge cells under full and faulty conditions of 25%, 50% and 75% of the submodules of each phase. In these conditions, the aim is to obtain the maximum energy transfer at the output terminals with the minimum harmonic content in the current and line voltage waveforms when using an internal PI control technique and the method of SPWM modulation (Sinusoidal Pulse Width Modulation) with multiple carriers, thus designing an isolated system for medium voltage. Two methods are addressed to compensate for faults in the submodules of the proposed multilevel modular converter. One of the compensation methods is based on the energy redistribution of the resilient submodules and the other one is based on an adaptation of the neutral shifting technique. The fault compensation method based on the energy redistribution of the resilient submodules restores 100% of the effects of the faults in the converter in all proposed conditions, both for currents and output voltages. The method for compensating faults using neutral shifting technique has no effect on compensating faults in the currents and phase voltages of the load, but it compensates 100% for the effects of faults under line voltage when they occur by up to 50% of the total of sub-modules and restores by 90% the effects of the line voltage fault when they occur in 75% of the total of sub-modules. Computational simulations with mathematically obtained specifications were performed on the Matlab/Simulink platform to validate the results.

The present Dissertation aims to demonstrate the development and construction of a DC-DC converter to harvesting energy ambient luminous indoor. The developed circuit it has the innovative characteristic that lets enjoy the low power indoor artificial light energy.The converter was created to work with input voltage less than 1 𝑉and without the use of an auxiliary battery, that is, only the ambient power source must be able to supply power to it and to the connected load at its output. To achieve this objective, several circuits were selected, including the starter, flyback, voltage lock circuit, UVLO and an oscillator. Throughout this work there is a detail of each chosen circuit as well as the reasons that guided its choice. The converter has as differential an ultra-low power microcontroller, that allows the search of the pointof maximum transfer of power of the source, in real time, using an MPPT technique. In this way, it is possible, in real time, to adjust the DC-DC converter to small variations in the light source so that it always extract the maximum power. All of this with low energy consumption, allowing the construction of a microcontrolled and autonomous DC-DC converter.

   From the various existing techniques to wirelessly transfer energy, one of the most efficient in terms of power and distance range is through magnetic resonance. In this technique, the transmitter and receiver coils are inductively coupled, and this coupling is intensified when both coils are tuned at the same resonant frequency. One way to enhance the power transfer efficiency and also to control the magnetic field direction is by applying metamaterial lenses which are able to overcome the diffraction limit. In this engineered material, the effective electromagnetic properties are functions of the chemical composition and geometrical arrangement of the unit
cells that compose it.

   The analysis of a resonant wireless power transfer with metamaterial lenses is a complex and computationally costly problem. Therefore, in this work, an alternative approach for the qualitative analysis of these structures is proposed by applying the thin-wire approximation and its numerical solution through the Method of Moments.
In addition, the Transformation Optics technique is applied to determine the refractive index profile of a metamaterial lens that leads to a desired magnetic field collimation.

   In this way, different magnetic metasurface lenses are designed, and their behaviour is computationally and experimentally verified. Finally, two-coil and four-coil resonant wireless power transfer systems are designed, built and tested with and without the inclusion of metamaterials lenses.

The use of numerical methods for the study and analysis of systems more complex energy transmission systems becoming more effective, obtaining results closer to those real conditions, due the technological development of computational systems. This research presents a numerical and computational modeling based on the Element-Free Galerkin Method (EFGM) evaluates and analyzes the magnetic fields , as well as the voltages and currents induced in the vicinity of electrical circuits and High Voltage Transmission Lines (TLs). The main characteristic of these methods is that they do not require a mesh , as it happens in the Finite Element Method, and is easier to work with different geometries, since the displacement of the nodes of the method for regions of interest is carried out with ease. There is only one set of nodes distributed in a region known as the problem domain. This method,is often used for solving partial differential equations. The time domain (TD) was used, which offers several benefits in electromagnetic problems, such as the transient field effect of an arbitrary time signal excitation, being that method EFGM is increasingly being used for eletromagnetic field computations and for modeling eletric devices. It has been shown that this method has a good convergence, sometimes better than that of the Finite Element Method (FEM).

Differential protection is the most common protection philosophy applied to power transformers. However, the conventional algorithms present operational inefficiencies, causing incorrect trip of the relays in non-fault situations. In this context, aiming to improve the traditional technique, this work presents a differential protection algorithm for power transformers based on Artificial Neural Network. The electrical system studied is modeled and simulated on the Alternative Transient Program software. The protection scheme is proposed for a 100 MVA transformer. The simulated electromagnetic phenomena are inrush, sympathetic inrush, overexcitation, current transformers saturation, and external and internal faults in the protection zone. The database created contains 1,210 files, being 80% for training and 20% for neural network testing. The developed algorithm has a specific module for the identification and detection of each simulated event. This computational intelligence tool is implemented on the MATLAB^® software. The proposed algorithm is also tested with an electrical system with an 18 MVA transformer. The obtained results show that the developed methodology presents superior performance when compared to conventional differential protection with harmonic restraint, proving to be a promising solution in this type of application.

Lightning are the main cause of unscheduled outages in transmission lines, in particular the backflashover phenomenon. In order to mitigate the risks associated with this phenomenon, it is important to model the elements of the system in a consistent way, in particular the tower-footing grounding and the transmission line. Thus, the present work has the objective of evaluating the influence of the tower-footing grounding system and the transmission line in the atmospheric overvoltages resulting from the direct incidence of discharges in overhead lines, using ATP / EMTP platforms.

These evaluations are carred out from simulations in the time domain and the impact of each representation is analyzed in terms of resulting overvoltages in the chain of insulators. The results associated with the simplified representations of grounding resistance at low frequencies and impulsive impedance are considered, as well as the results from the rigorous representation of the dependent behaviour of the grounding frequency. In addition, two representations of transmission lines are evaluated: the consecrated J. Marti and the model named in this work as modified Marti’s model that unlike the J. Marti, considers the variation of the parameters of the soil along with the frequency.

According to the results, the grounding representation has a significant impact in the resulting overvoltages and in the definition of the critical value of the lightning current that will lead to a fashover. Considering the representation via harmonic impedance as reference, the impulsive impedance leads to critical current values very close to this rigorous representation and can be seen as a good option for the grounding modeling in transmission lines for the study of their performance in ATP/EMTP platform.

Still according to the results, line models have less influence on the resulting overvoltages in the chain of insulators. Only in high resistivity soils the effect of including the dependence of the electrical parameters of the soil with the frequency in the modified Marti’s model leads to perceptible differences in the results overvoltages developed, especially when considering the midspan incidence.

Power transformers are essential equipment for the supply of electrical energy, whose repair or replacement implies longer interruption time, direct and indirect costs. The high cost of this equipment and its importance for the electrical system requires special attention to its operation, protection and maintenance. It is essential that electric power companies invest in protection systems that guarantee safety, reliability and higher quality of service to the electrical system. The main methodology adopted for the protection of power transformers corresponds to the differential protection, using traditionally the estimation of fundamental frequency phasors to
calculate the operating and restriction currents. From an operational plan, the operation of the relay is mapped in regions of actuation and blocking (normal operation). In order to discriminate faults and energization conditions, the harmonic analysis by the Fourier transform is normally used. Despite the great applicability of the differential protection using the tool proposed by Fourier, the differential relay can be sensitized by transient events in which the protection is not desired to isolate the system. The main limitation of the Fourier method
consists in the inability to analyze the differential currents of operation and restriction in timefrequency. The present work, therefore, proposes the use of the wavelet transform whose main characteristic is to analyze signals in both the time domain and the frequency domain. From the multi-resolution analysis technique (MRA) of the wavelet transform, the analysis signal is decomposed into approximation and detail coefficients and, through the spectral energy variation of the coefficients, an operational logic is proposed to discriminate faults other events. The efficiency of the algorithm is attested from the submission of the algorithm to a set of
simulations obtained from the electrical system modeled in the ATPDraw software with enough elements to reproduce the effects of the electromagnetic transients in the power transformer.

This dissertation deals with the robust H ∞ control synthesis by static or dynamic output feedback for time-invariant linear systems, discrete time, with dependency on parameters. The robust control synthesis is problem of difficult solution, since it requires solving a semi-infinite optimization problem, that is, finding an optimal controller that minimizes the worst case performance and guarantees the stability of the system for the infinite domain of uncertainty. The combination of polytopic models and formulations based on linear matrix inequalities for analysis and synthesis of robust control systems is very popular, since it is possible to transform the problem of difficult solution into a simpler problem, verifying only the vertices of the polytope. However, there are restrictions on the type of problems that can solve, the computational cost can increase quickly with the number of vertices of the polytope, making the procedure computationally infeasible, and, in some cases, it is not possible to find the best solution or even a controller robustly stable. In this dissertation, it is proposed to apply a strategy different from usual to solve the problem of semi-infinite optimization considering an iterative two-step procedure, synthesis and analysis, in which each stage has problems of optimization easier to be solved. In both steps, synthesis and analysis, the differential evolution method is used, an evolutionary optimization algorithm of simple implementation, with few adjustment parameters, making it possible to find efficient solutions for various optimization problems, covering multimodal problems. The presented procedure can be applied to any type of control problem considering controllers with any structure and order. Illustrative examples will be presented applying the proposed procedure and others used in the literature for comparison, where in the presented method results in better solutions in some of the examples, the only method being able to provide a robustly stable controller for the latter example considering the methods available in the literature up to the present moment, proving the efficiency of the robust control synthesis procedure.

DC-DC converters are used in various industrial and residential applications. One of them is in inverters of photovoltaic systems. The DC-DC converter is used to control the voltage of the photovoltaic string to a desired voltage level on DC-AC converter. Generally, it is used a boost step-up converter. The purpose of this work is to analyze the use of a DC-DC up-down converter of SEPIC topology, as well as to develop its modeling and control design in current mode. The purpose of using the current mode control is to eliminate the photovoltaic panel interface capacitor. The SEPIC is a fourth-order converter (four-element energy storage) and it's control design can be complex. The design control methodology presented in this work uses the module condition of the root locus method. Maximum Power Point Tracker (MPPT) algorithms are used to extract maximum power of photovoltaic systems. MPPT algorithms are effcient for voltage-mode control, however because of the current saturation problem in the face of falling solar irradiance, they lose efficiency. The MPPT proposed in this work controls the photovoltaic panel current and acts on current saturation problem to increase the efficiency of the system. The simulated results for a tree-phase 1.6 kVA showed the proposed MPPT dynamic e efficiency.

The distribution and transmission systems supply a huge variety of loads. Nevertheless,
most of these loads are non-resistive or fluctuating. These loads can generate voltage
variations which affect the operation and efficiency of the systems. In order to minimize
these problems, reactive power compensation is indicated. In this context, the Static
Synchronous Compensator (STATCOM) has been widely used. The major challenge in the
design of medium and high voltage STATCOMs is to define a converter topology which
must reach higher power and voltage levels with standard rated semiconductor switches.
In this context, the topology Double-Star Chopper Cell (DSCC), member of the Modular
Multilevel Cascade Converter family, has proved to be an attractive option for medium and
high voltage STATCOM applications. Nevertheless, the DSCC topology has some challenges
in the modulation schemes, energy storage requirements and reliability. Regarding the
energy storage requirement the literature considers balanced capacitor voltages in the design.
However, depending on the modulation strategy, the capacitor voltages oscillate within a
range and with a spreading. In terms of reliability, the literature presents works in which is
evaluated the DSCC lifetime. Nevertheless, none of these works take into account the effect
of the non-negligible voltage ripple in the cell capacitors and also how the modulation
strategy can affect this capacitor voltage ripple and, consequently, the capacitor lifetime.
Therefore, this work considers the effect of all cell components (i.e. semiconductor devices
and capacitors) and analyzes the modulation strategies impact on the energy storage
requirements and reliability of a DSCC-STATCOM. For this purpose, two modulation
strategies are selected: Phase-Shifted Pulse-Width Modulation (PS-PWM) and NearestLevel Control with Cell Tolerance Band algorithm (NLC-CTB). In addition, this work
introduces the spreading factor index to compare different modulation strategies in terms
of capacitor voltage balancing capability. Additionally, the energy storage requirements
for each modulation strategy are discussed. In order to analyze the cell capacitor voltage
ripple effect on lifetime, the methodology which better represent its effect on the capacitor
lifetime model as well as its sampling frequency are investigated. Moreover, the results
indicated that different modulation strategies may require distinct energy storage. In
fact, the NLC-CTB requires energy storage 44 % higher than PS-PWM. As consequence,
different capacitor numbers impact on the lifetime of the capacitor bank. On the other
hand, the modulation strategies may produce different power losses which also impact on
the lifetime and produce additional costs.

Among the existing industrial processes, a particular case is those that present in their dynamics high delay, or dead time. For this case, previous studies in the literature indicates that classical techniques employing only PID controllers have difficulties in meeting the stability and performance criteria. A solution to this problem is known as Smith predictor that allows designing the controller for the model free of delays. However, it is a scheme sensitive to the uncertainties of plant parameters. In this dissertation, an iterative tuning procedure for PID controller with a filtered Smith predictor is applied for multivariable, with number of inputs equal to the number of outputs or not, continuous time, time-delayed systems, with uncertain delays varying in ranges. The most important contribution of this work is to present an alternative methodology to treat the problem of multivariable controller tuning in uncertain systems with dominant delays. The proposal is to tune the PID control with Smith predictor and robustness filter so that the closed loop outputs of the controlled plant approach the outputs of a reference model, both subject to the same exogenous inputs. Such a control objective can be formulated as an optimization problem where it is desired to minimize the integral of the squared error, or other criterion. A diagonal reference model is adopted to enforce decoupling among the control loops. Each diagonal element of the reference model determines the characteristics desired for the reference tracking responses. When considering that part of the system parameters are not precisely known, but that they may be within a closed interval, the multiobjective optimization problem becomes a multiobjective semi-infinite optimization problem, difficult to solve, where it is necessary to minimize the worst case in the uncertainty domain with infinite points, for both the reference signal and the disturbances. To solve this problem, it is proposed to apply an iterative two-step procedure. In the first step, the optimization is transformed into mono-objective and performed for a finite set of points and, in the second step, the achieved controller is verified for the points of the infinite uncertainty domain. If necessary, new points may be added to the set until the result of the optimization in the first step is similar to the result of the analysis in the second step. In this work, the Differential Evolution algorithm is applied to both the tuning step and the analysis step. Some illustrative examples are considered to demonstrate the efficacy of this technique and the results obtained by the proposed procedure are compared to others in the literature.

In this work, state feedback controller synthesis conditions are proposed for discrete-time uncertain systems with delayed states and saturating inputs. Actuator saturation is represented in terms of a nonlinearity of the dead zone type, with the consequent application of the generalized sector condition. Due to saturation it is necessary to estimate a region of attraction within which all trajectories initiated within it remain in that region and converge to the origin. Thus, the objective is to determine conditions for the synthesis of state feedback controllers that guarantee for a set of allowable initial conditions the asymptotic stability of discrete time uncertain systems with state delay and actuator saturation. To this end, two approaches are proposed: the first using a suitable Lyapunov-Krasovskii function that treats delayed states and the second rewriting the delay as an augmented system free of delay and swicthed by the delay value. Thus, the main contributions of this work are three. The first is to propose a sum inequality responsible for varying delays in the time when a method is provided to deduce a less conservative stabilization condition. The second is to propose a new characterization of the allowable set of initial sequences that consists of the definition of three sets, producing larger initial sequences than similar approaches in the literature. Lastly, it was verified that the inclusion of the extra signal composition of the generalized sector condition can increase the size of the region of attraction estimate. And finally, numerical examples are presented to demonstrate the efficiency of the proposed methods and to compare with others in the literature.

Medium/high-voltage distribution and transmission systems have a huge variety of nonlinear loads. Additionally, the large penetration of renewable energy sources in the power system has motivated studies about static synchronous compensator (STATCOM). In this context, the multilevel converter is an interesting technology for this application. Particular attention is given to modular multilevel cascaded converters (MMCCs) due to the high scalability, modularity, redundancy and efficiency that are inherent features of the topology. To be able to operate at high voltage levels, a large number of semiconductor cells/devices is required, compromising the converter reliability. Redundant cells can be employed to fulfill the converter reliability requirements. However, the redundancy factor design is still an important area to be explored. This work proposes a reliability model for the correct selection of the redundant cells number, which aims to reach a determined lifetime target. This model combines random failure and wear-out failure analysis. The redundancy factor is computed for given reliability requirements, i.e., this work proposes a reliability-oriented redundancy design. The case study is based on a 13.8 kV/17 MVA MMCC-based STATCOM. The implementation of the converter is performed through four commercial blocking voltage of semiconductor devices (1.7, 3.3, 4.5 and 6.5 kV). The capital expenditure (CAPEX) and operation expenditure (OPEX) are estimated considering the required redundancy factor. Considering a target of 20 years of operation for the converter and reliability levels greater than 90%, the design based on 1.7 kV devices has the lowest cost for the MMCC STATCOM realization.

The synthesis of multivariable PI controllers by means of optimization is aneffective strategy that has been widely used thanks to the variety ofoptimization methods and its great capacity of finding good solutions. Thisprocedure has two fundamental steps, the modeling of the problem to beoptimized and the choice of the optimization method to be applied. Theobjective function has direct influence over the computational cost and onthe characteristics of the closed-loop system. The algorithm to be utilized,by its turn, impacts on the capacity of finding or not good solutions with anacceptable computational cost. This work presents a variant of theDifferential Evolution method which takes advantage of artificial neuralnetworks to reduce the computational cost of the synthesis process formultivariable PI controllers. The new algorithm is tested with classicalbenchmark functions of the optimization area for its validation and study.Two typical systems of the control area, both with multiple inputs andoutputs and with transport delays are used as examples of the applicationof the new algorithm to the synthesis of multivariable PI controllers. Theresults show the benefits and drawbacks of the proposed method. Besidesthat, as a secondary objective, the differences of using objective functionsbased on systems and signals norms are analysed and discussed. Theproposed method reduces the total number of objective functionevaluations, decreasing the computational cost. However, it was observedan increase in the results variance and a decrease in convergence rate.The objective functions based on signal norms present a better controlperformance, while objective functions based on system norms have alower computational cost.

With the technological development, the use of wireless systems has become common usual and a great number of electromagnetic waves have been scattered in the environment. A large part of these waves does not achieve the receiver and become unusable, create energy waste. Thus, the rectenna arise how a promising system to radiofrequency energy harvesting and conversion into electric energy of direct current. Since the rectenna is composed of an antenna associated with a rectifier, the antenna design is of utmost importance to obtain a system with high efficiency in the conversion of harmonic signals into direct current.
In this context, the aim of this work is to design a printed antenna with rectangular patch, fed by a microstrip to integrate a system rectenna. For this, an optimization was carried out employing the Differential Evolution Algorithm, which has been applied in the resolution of problems involving antennas, presenting better performance than other classic evolutionary algorithms. The selection step in this algorithm was modified, based on the NSGA-II algorithm for the treatment of the multi-objectives.
For calculating of the objective function it is necessary to use computational methods that analyze the fields around the antenna continuously, whenever the antenna geometry is modified. Thus, it can be observed that the Finite-Difference Time-Domain Method is appropriate for the simulation of the proposed antenna, besides carrying out the analysis in the time and for several frequencies in a single simulation. The association of this method with an absorbing boundary condition simulates an unlimited space, absorbing the fields that reach the boundary of the computational domain, thus ensuring the accuracy of the results.
In the design the antenna dimensions - patch width and length and inset-fed length - are used as optimization variables to reach the goals defined for the antenna design - resonance frequency of 2,45 𝐺𝐻𝑧, bandwidth in the range of 2,4−2,5 𝐺𝐻𝑧, return loss less than −15 𝑑𝐵, resistance of 50 Ω and zero reactance. The optimization goals were reached for the input variables used and the simulation results showed agreement with the answers given by commercial softwares.

This work presents a study of the Element-Free Galerkin-free meshed method applied in solving low-frequency electrical grounding problems. The electromagnetic model of the grounding system is developed from the application of the current injection technique directly at the defect point as a way of excitation the grounding electrode. Initially, the main characteristics of meshless method and its general concepts are presented. Then, the formulation of the problem to calculate the ground resistance and determination of the potential distribution for an electrode inserted in homogeneous soil will be developed, as well as in a heterogeneous soil stratified in two and three layers. For the implementation of the electromagnetic model, a computational tool was developed using the MATLAB environment. The results obtained through the application of the meshless method are compared with values generated by MoM, demonstrating the validity and precision of the developed technique.

In the last years rectennas technology has received increasing attention of researchers due to the extension of the possibilities of its application in the supply of energy for the current loads with very low level of energy consumption. In this context, this work presents a theoretical and experimental investigation about the use of rectenna technology with the objective of electrically charging a battery model widely used in electronic devices, the Li-Ion battery. Initially a broad bibliographic review on the subject is carried out to support the investigations conducted in this work. The rectenna modeling process was started by optimizing the rectifier with the load using the ADS® software. Different prototypes of the rectifier circuit were manufactured, operating at the frequencies of 2.45 and 1.80 GHz and a circuit capable of operating at both frequencies simultaneously. A DC-DC booster converter circuit has also been investigated in order to enable rectenna operation from low input power levels. In this work, printed antennas were used, simulated in the software CST®(Computer Simulation Technology) and connected to the rectifier. The dimensions of the antennas were optimized in order to obtain values ??of parameter S11 below -10 dB and the highest possible directivity in order to allow a larger collection of the electromagnetic energy. Two electromagnetic lenses were designed and manufactured to compose the rectenna, based on metamaterial technology, in order to promote a superior collection of energy. Finally, the rectennas were subjected to operational tests of charge of the Lithium Ion battery. The rectennas were connected in parallel in the DC of the circuits in order to obtain the maximum supply of current for the battery.

Photovoltaic energy has stood out in the current context of energy generation. The fundamental element of a photovoltaic system connected to a electric grid is the electronic converter. In order to be marketed, the electronic converters must be tested, certified according to current standards and receive the INMETRO seal of compliance. In the certification process, an equipment known as emulator of photovoltaic modules is used. This work aims to present the modeling and controlling structures of three topologies of emulators of photovoltaic generator, they are, emulator based on circuit of Thevenins, converter buck with bus constant cc and in rectifier PWM and converter buck. In addition, realize a performance analysis of the topologies connected to a photovoltaic inverter, in different environmental conditions and transient response; analysis of the MPPT dynamic and instantaneous efficiency and the loss and efficiency analysis of each topology. The photovoltaic emulators studied in this work have the capacity to test single-phase and three-phase inverters with a power of 10 kW. This value was defined based on Ordinance n° 004 of January 4, 2011, this power range covers the majority of the inverters sold for residential applications. The results show that the photovoltaic generator topologies presented are able to emulate the behavior of a photovoltaic array during solar irradiation and temperature variations. The topology based on buck converter presents the highest efficiency. However, the use of each topology will depend on the operator's objective once each one of them has both positive and negative points for different applications.

Uninterruptible Power Supplies (UPS) are equipment used to provide continuous and high quality power in critical systems such as life support systems and transportation systems. UPS increase the reliability of power supply to levels higher than those achieved in utility networks. However, like all electronic equipment, UPS are subject to failures. Thus, reliability can be enhanced by applying redundancy when performing a UPS connection in parallel. The present work focuses on the study of the stability enhancement and simplification of the small signals model to the parallelism of UPS without communication, that is, it uses just local variables for power control. Each of the subsystems present in the implementation of parallelism without communication using droop method is analyzed in detail. The impact on eigenvalues and damping was investigated using different cut-off frequencies in the active and reactive power measurement filters in the presence and absence of a Power System Stabilizer (PSS). The proposed techniques are validated through simulations and lab tests.

This work presents the comparative study regarding two nonisolated dc-dc converters for high-power, high-current applications, i.e., the three-phase interleaved buck converter and the buck converter based on the four-state switching cell. The operation in continuous conduction mode is analyzed for an operating point where the duty cycle is between 2/3 and 1. The qualitative and quantitative analyzes are properly developed, thus allowing the accurate design of the power stage elements, which include the filter inductor(s), output filter capacitor, switches, and diodes. The design is validated by simulation, where waveforms are obtained for ideal and nonideal circuits considering real components defined in the design procedure. The results are analyzed in detail, aiming to identify the main advantages and limitations of the topologies. As general characteristics of the studied converters, it can be stated that the operating frequency of the filter elements is three times higher than the switching frequency, as the current through the semiconductors is one third of that regarding the elements of the classical converters. Thus it is possible to achieve significant reduction of size, weight, and volume, as well as improved dynamic response.

The purpose of this dissertation is to present a performance evaluation of the protection relay distance function when applied in series compensated transmission lines. This type of protection has its operation based on line impedance, so it is affected by problems related to series compensation. The main operating characteristics of this type of compensation are presented with greater emphasis on the controlled system (TCSC). It’s also mentioned the most common effects of series compensation in the operation of the distance protection. These effects are not only related to the change in impedance of the protected line, but also with other phenomena, such as transients and the operation of capacitor protection systems. For the proposed evaluation a practical transmission line assembly was developed with reduced voltage, as well as a three-phase TCSC module that allows to compensate the line by up to 40%. There are described tests in which different types of faults with different compensation characteristics were simulated at a specific point in the proposed system. During the tests, the voltages and currents of the system are sampled by means of a data acquisition board and later processed in an algorithm, developed in Matlab software, which simulates the MHO characteristic distance protection function of a relay. A simulation with similar characteristics to the practical approach was developed in Simulink software to validate the results. Finally, the conclusions of the analysis are presented.

The mathematical modeling of engineering problems often involves the use of one or more algebraic, differential or integral equations. However, it is possible to determine an analytical solution only for a small part of these problems. In most of the cases, it is an extremely complex or even impossible proccess. In these cases it is necessary to use computational numerical techniques in order to obtain approximated solutions for such problems. An important class of numerical techniques is the class of the meshless methods. These methods, differently from finite element or finite difference methods, does not rely on meshes. For the discrete problem domain representation they use a cloud of nodes scattered all over the domain and its boundaries, not requiring any relation to be stablished between the nodes.

Meshless methods have already been successfully employed in the solution of electromagnetism problems, both in low and high frequencies, including wave propagation problems. Propagation problems like electromagnetic scattering are in general open or unlimited problems and its numerical solution demands the employement of some domain limitation thecnique. Absorbing boundary conditions, that were traditionally used to limit domains, have been replaced by newer thecniques based on the perfect matched layers concept, the PML. Among the several existing PML variations, stands out the Uniaxial PML (UPML), which was developed from the physical interpretation of the PML originally proposed concept. According to this approach the domain is surrounded by layers of an anisotropic fictious material whose characteristics are specially designed so the layers acts as an efficient absorber perfectly matched to the domain medium. The UPML limits the domain yet simulating the unbounded wave propagation.

This work investigates the bidimentional electromagnetic scattering problem, its definitions and mathematical modeling. It also analyses the IEFG (interpolating element-free Galerkin) meshless method, its characteristics, mathematical background and formulation. The plane wave scattering by a z-infinite conducting circular cylinder is solved by the combined IEFG-UPML numerical technique. Analysis of the proposed technique is firstly done in the form of a parametric analysis of the IEFG and UPML methods. Then, the method performance and the accuracy of the solutions generated by the new parameters set are evaluated while solving the scattering by cylinders of bigger dimensions. The analytical solution and the numerical solutions generated by the IEFG-ABC method – with a first order Bayliss-Turkel absorbing condition – are used in this comparative analysis.

Since the invention of the wheel, the man wish to find something that can be move alone. There were three types of vehicle technology in the 19th century: steam car, combustion car and electric car. But the combustion car became the predominant technology due to the discovery of huge oil fields in Texas, US. However, the concern about environmental pollution has contributed to the return of electric vehicle, as an alternative to reduce the greenhouse effect. Therefore, low autonomy and the time it takes to recharge remain as the great disadvantages of electric vehicles. The most advanced battery used in electric car is the Li-ion battery. The battery is used together with ultra-capacitors to contribute with the increase of performance of car and extend the life time of battery. Regenerative braking system is the other technology present in electric vehicle. It converts the braking energy in electric energy that it is transferred to the bank of batteries and it contributes to extend the autonomy of the electric vehicle. Every electric vehicle is a plug-in car because it requires to connect to grid to recharge the battery. Therefore, most of electronic devices can cause distribution circuit congestion and affect the power quality. The study presents the behavior of main electric parameters during the recharge of the electric vehicle Fiat 500e. The analysis of results were done based on the limits recommended by national and internationals standards about quality of energy. In sequence, these parameters were included on load curve of Brazilian consumers in different scenarios of recharge. The results contribute to discuss the general effect on distribution system caused by the spread of electric vehicle and how the recharge in different time of the day can affect the consumption and consequently the Power System.

The Modern power transmission networks are extensive and complex systems that have been experienced great blackouts over past two decades. These rare but significant events raise concerns about stability and safety of the electrical system operation. Segmentation of large AC systems through DC links introduces a new concept that utilizes the advantages of direct current transmission to improve grid reliability and increase power transfer capacity. Technical literature argues that the segmentation of the AC network and the introduction of DC links at these systems connection points bring benefits to system operation, once contingencies generated on one side of the DC connection point would not be reflected on the other side of DC connection, thereby reducing the likelihood of cascading outages and blackouts due to load restraint on transmission lines and transformers. Although the technical literature defends the advantages DC links utilization for grid segmentation, this subject stills in constant debate by scientific community, because the number of existing DC links in operation are insignificant, generating doubts about its effectiveness for power flow control at contingency cases. Amidst this scenario, this paper present a study of the main topics regarding the use of this new network segmentation philosophy, bringing a practical point of view for the use of this concept at the electrical power system planning. The effect of DC segmentation before a contingency that would initiate major outages in an adapted electrical system model IEEE-14 bus is studied and simulations have been performed with test HVAC systems and segmented by HVDC link. The results has been compared, principally in relation to the voltage bus, reactive power generation, system losses and power flow at the lines, and demonstrated that this new concept improved the grid confiability.

Microstrip antenas in general are used in automotive, aeronautics and aerospace applications in which small size and weight and cost reductions are required, high performance, installation readiness and aerodynamic profile. For the antennas design, computational software are needed to verify their behavior in such frequencies band. These software allow the simulation of different antennas geometry with material variation with no need of prototypes assembly, which reduces project time and costs. There are many numeric methods to simulate microstrip antennas, highlighting MoM, FEM and FDTD. In this research it was chosen FDTD between those, due to the designed antennas owns complex geometry to operate in ultra wide band and FR-4 fiberglass dielectric which is a high loss material and this method obtains good performance for the antenna simulation of such configuration. To validate the FDTD code, it was first simulated an antenna with simpler geometry. Once the code was validated, it was tested the influences of general settings in the code for the antenna simulation result, like time step, antenna geometry, excitation and measurement points definition and time step to release the excitation pulse. Since the chosen antenna in the research has complex geometry and high loss dielectric material it was necessary to investigate the excitation source more appropriate for the problem, among these: voltage resistive source, magnetic wall and incident wave subtraction from the total wave. To compare the simulated results it was performed a simulation of the same antenna geometry in a commercial software. The simulation results of each of the excitation sources investigated demonstrated good matching when compared to the curve generated in the commercial software, highlighting the magnetic wall. The antenna developed was assembled after the simulation validation. It was used a fiberglass PCB with double face with FR-4 dielectric to assembly the antenna with the same dimensions used in the simulation. To avoid noise in the return loss measurement the antenna was inserted in an anecoic chamber. There was a good matching between the simulations performed in the FDTD software developed in the research, the commercial software and the antenna assembled measurements.

It is common for industrial processes to be multivariable and therefore there is the need to design and implement control systems for these processes. It is well known that not all multivariable processes have the same number of inputs and outputs. Systems in which the number of input variables is different from the number of output variables are known as nonsquared systems. For multivariable processes with high interactions among the control loops or that require more stringent control, control systems with decouplers are often employed. Initially, it is seeking to eliminate the undesirable effects of the interactions among the control loops by means of the decouplers to project the individual controllers. In this work, we will investigate the use of a nonlinear optimization algorithm for the synthesis of PI control systems with decoupling, considering multiple control objectives, to be applied to non-square multivariable systems with multiple time delays. The main strategy of the syntheses is methodology is to consider as an optimization function the approximation of a closed-loop reference model to attain the transient response specifications of the multiple control loops as well as the decoupling between them. In addition, other functions may be included, according to the problem being treated, related to control effort, disturbance rejection and/or measurement noise attenuation. We adopt in this work the Differential Evolution algorithm to achieve the solution of the non-linear multiobjective optimization problem. It will be demonstrated, through examples, that the proposed methodology for synthesis of control PI with decoupling for non-squared multivariable systems may present better results than other methods already published to deal with the same type of problem.

Renewable energy sources have increased their share in the global energy matrix. In this context, photovoltaic solar energy has shown to be promising due to the fall in the prices of the system as a whole and also due to the abundance of the solar resource. Brazil, despite the high level of solar irradiation, is still underdeveloped in this type of source of electricity generation compared to other countries with lower radiation levels such as Germany, Italy and Spain. It is true that the countries that currently have a large share of this type of source have an electric matrix with a large share of sources of fossil fuels, which explains the high level of incentives practiced with a view to making renewable sources viable. Brazil, in turn, has implemented regulatory and tax measures in order to promote distributed generation and thereby increase the share of wind and photovoltaic sources in its already predominantly renewable matrix. This work has the objective of studying and analyzing the economic viability of the implementation of photovoltaic systems for consumers of the residential class connected to the 63 distributors of electric energy that make up the Brazilian electrical distribution system. The methodological procedure that led to this research were the studies based on indicators of investment analysis and the calculation of the level of energy cost - LCoE. As a conclusion, this research demonstrates that tariff parity is already a reality for photovoltaic energy when compared to energy tariffs for residential consumers.

Lately, in the electrical sector there have been technological developments that contribute to its modernization, safety and quality. All of the automation, data collection and processing throughout the system, from the generation, transmission and energy distribution, has been worked within the concept of Smart Grid, in which there is the emergence of new equipment, for protection, control and measurement power. Smart metering is being developed to fulfill this new concept of energy distribution, not only measuring the energy supplied by utilities to the consumer, but also that generated by the consumer and supplied to the system assuming a bidirectional flow of energy. The number of data reproduced by these meters and stored in the database will increase considerably after the scale installation. These data contain relevant information that, even though they are already stocked, are still not processed, due to lack of studies and research in this area. This work proposes a methodology for the data processing, using computational intelligence tools, in order to extract relevant information for the development of new services for consumers and power generators. It presents a methodology for evaluation the energy consumption and the demand curve of a public building, aiming to assist the consumer, the concessionaire and the energy sector in the level of planning and future decisions. To exemplify and validate the proposed methodology, the computational intelligence tools were used. The Fuzzy Logic and the Artificial Neural Networks are also described in this work, showing their characteristics, configurations and comparisons between performances of the different algorithms used. Promising results were achieved with the methodology described, proving the efficiency of the methods and algorithms presented in this work and also serving as a basis for future studies in this area of energy mensuration used by different types of consumers.

The photovoltaic systems characterization is an emerging issue that has attracted the attention of many researchers recently, since photovoltaic generation is expanding worldwide. The characterization of photovoltaic systems under real operating conditions is still a challenge. The high costs of the characterization systems and the difficulty of measuring the several variables involved in the process make it difficult to characterize them on site. In general, photovoltaic panels are characterized by manufacturers by laboratory tests under Standard Test Conditions (STC). These Standard Test Conditions are determined for a temperature of 25C, irradiance of 1000W=m2 and air mass (AM) of 1:5G established by IEC-60904-3:2016. However, these conditions are hardly to found under real working conditions of photovoltaic panels because they are subject to various types of interference, for example: shading, reflection and temperature variations. Hence, the development of low-cost methodologies and systems that allow the characterization and analysis of panels in real operating conditions becomes of great relevance. In this work a measurement and characterization system were developed using of low-cost sensors. Several experiments were carried out, using a photovoltaic panel of 80W, on different days and hours to provide a database for analysis. From the collected data of relevant electrical and environmental variables, it was possible to obtain the experimental characteristic curves of the panel under different operating conditions. In addition to the experimental data, computational simulations were made to obtain the characteristic curves based on the two-diode model. The experimental and simulated characteristic curves were compared in the analyzes. It could be observed that for some ranges of incident irradiance and temperature the model predicts a lower power generation than the obtained-on field. It is presented analyzes related to the impact of the air mass on the distribution of the irradiance for the bands of the evaluated solar spectrum (UV, visible and global) besides the influence of temperature. The results obtained can be used to yield more accurate models and demonstrate the importance of performing the characterization of the modules under real operating conditions.

Within the wide range of numerical methods applied to the solution of Partial Differential Equations are the Meshless Methods. What characterizes this class of methods is the fact that they do not have meshes used as in Finite Element Methods. When it comes to structures that constantly changes their shape, this type of approach can be very efficient, reducing the cost of rebuilding the mesh at each iteration. However, such solution has a relatively high computational cost for numerical resolution. This work presents parallelization techniques applied to problems in electromagnetics by Meshless Methods. Throughout this paper, parallelism techniques through shared memory and message-passing are presented, and the use of parallel solvers used in the solution of linear systems. The problem used as a case study is the model without mesh of a three-phase induction electric machine in motion, which can be evaluated in the magnetic flux in its interior, the currents and voltages in their drivers, as well as possible to see their movement.