J. O. P. Pinto

Field Programmable Gate Array (FPGA) Based Neural Network Implementation of Stator Flux Oriented Vector Control of Induction Motor Drive

In this work, it is proposed the implementation of the SFOVC-ANN using field programmable gate array (FPGA). The proposed scheme assure parallel processing of the ANN, since the circuit design was done in such way that the neurons in the same layer processes the input signals in parallel. The non-linear sigmoidal transfer function is implemented using Spline Interpolation, which guarantees an excellent precision. Initially in the digest, a description of the proposed system is given. Then, the nonlinear neuron implementation strategy is explained. Following this, the FPGA implementation of ANN is described. Finally, simulation and experimental results are given to substantiate the development.

Controle ótimo H∞ de sistemas não-lineares com modelos fuzzy takagi-sugeno

A design method for tracking system with disturbance rejection applied to nonlinear systems using fuzzy control is proposed in this paper. Fuzzy feedforward controllers M(a) and N(a) are designed in order to obtain the tracking system. These controllers minimize the H∞-norm from the reference input signal r(t) to the tracking error signal e(t), where the tracking error signal is the difference between the reference input signal r(t) and the output signal z(t). A dynamic feedback fuzzy controller Kc(a) is designed in order to minimize the H∞-norm from the disturbance input w(t) to the output z(t). The designs are formulated in Linear Matrix Inequality (LMI) framework, such that the optimal solutions of the stated control problems are obtained. Simulation of a practical problem illustrates the effectiveness of the proposed method.

Fraud detection system for high and low voltage electricity consumers based on data mining

This work presents two computational system for fraud detection for both high and low voltage electrical energy consumers based on data mining. This two kinds of consumers demanded different approaches and methodologies. The first is based on SOM (Self-Organizing Maps), which allows the identification of the consumption profile historically registered for a consumer, and its comparison with present behavior. The second is based on a hybrid of data mining techniques. From the consumer behavior pre-analysis, electrical energy companies will better direct its inspections and will reach higher rates of correctness. The validation and results showed that the two systems are efficient on the cases of lower consumption resulted by fraud.

Real-time selective harmonic minimization in cascaded multilevel inverters with varying DC sources

A new approach for selective harmonic elimination in a 7-level cascaded multilevel inverter with separate DC sources will be presented. As opposed to previous research in this area, the DC sources feeding the multilevel inverter are considered to be varying in time. This method uses genetic algorithms to obtain switching angles offline for different DC source values and uses neural networks to determine the switching angles that correspond to the real-time values of the DC sources. This implies that each one of the DC sources of this topology can have different values at any time but the output fundamental voltage will stay constant and the harmonic will still meet the specifications. The paper gives details on the approach used, together with simulation and experimental results.

A novel power calculation method for droop-control microgrid systems

In this paper a new methodology is presented to calculate the average power quickly and accurately for the single-phase paralleled inverters intended to be applied in a droop-control microgrid system. Most existing droop control systems utilize a simple first-order filter to calculate the active and reactive power. The added filter can smooth the calculated results but it tends to hurt the system dynamic response due to a low cut-off frequency that is intended to get enough ripple attenuation. This low frequency pole in the low-pass filter could also introduce system instability under certain load conditions. A new filter calculation methodology is thus proposed to achieve better ripple attenuation with faster response by combining a low-pass filter, a delay, and summation blocks. Both simulation and experiment are used to verify the effectiveness of the proposed power calculation methodology.

Multiple signal processing techniques based power quality disturbance detection, classification, and diagnostic software

This work presents the development steps of the software PQMON, which targets power quality analysis applications. The software detects and classifies electric system disturbances. Furthermore, it also makes diagnostics about what is causing such disturbances and suggests line of actions to mitigate them. Among the disturbances that can be detected and analyzed by this software are: harmonics, sag, swell and transients. PQMON is based on multiple signal processing techniques. Wavelet transform is used to detect the occurrence of the disturbances. The techniques used to do such feature extraction are: fast Fourier transform, discrete Fourier transform, periodogram, and statistics. Adaptive artificial neural network is also used due to its robustness in extracting features such as fundamental frequency and harmonic amplitudes. The probable causes of the disturbances are contained in a database, and their association to each disturbance is made through a cause-effect relationship algorithm, which is used to diagnose. The software also allows the users to include information about the equipments installed in the system under analysis, resulting in the direct nomination of any installed equipment during the diagnostic phase. In order to prove the effectiveness of software, simulated and real signals were analyzed by PQMON showing its excellent performance.

Voltage estimation in electrical distribution systems

The increasing demand for high power quality has increased the demand for power quality monitoring tools. The voltage performance monitoring for each feeder is one of the needs found by the utility companies. Since it is not economically viable to measure every single node in the system, it is necessary to use estimation techniques in order to get all needed information with a reduced the number of meters. These are basically interpolation techniques. Each interpolation provides different performance of estimations accordingly the application. In this work, it is proposed a non-linear non-parametric method which was found to get the best voltage estimates of feeders in relation to other more usual techniques. Comparative results for different methodologies in a hypothetic system are presented and discussed.

Dynamic adaptation of droop control curves for microgrid connected inverters with variable input power

In this work a strategy for response dynamic optimization of a microgrid composed of an infinite bus (utility) and two inverters is proposed. Through the differential evolution technique, the optimized parameters of the droop control characteristics for operation in several equilibrium points are obtained. With an artificial neural network, these parameters are corrected for each operating point in order to ensure the stability as well as reduced settling time and damped response free of overshoot. The stability after changes in the operation point and the generalized dynamic response optimization are evaluated through simulation results using the software Matlab/Simulink® and experimental results of the proposed microgrid.

Power flow optimization for grid connected inverter using evolutionary algorithm and additional control loop

In this work it is proposed to validate an evolutionary tuning algorithm in plants composed by a grid connected inverter. The optimization aims the tuning of the slopes of P-ω and Q-V curves so that the system is stable, damped and minimum settling time. Simulation and experimental results are presented to prove the feasibility of the proposed approach. However, experimental results demonstrate a compromising effect of grid frequency oscillations in the active power transferring. In addition, it was proposed an additional loop to compensate this effect ensuring a constant active power flow.

Unified Hybrid Power Quality Conditioner (UHPQC)

This paper proposes the Unified Hybrid Power Quality Conditioner (UHPQC). The proposed topology manipulates, at the same time and in continuous way, the displacement power factor and also it mitigates the dominant harmonic component of the line current. The structure is composed by a small rating Active Power Filter (APF) in series with a single passive filter. The main objective of this topology is to control the reactive power injected by the passive filter into the system by controlling the voltage applied on its terminals and simultaneously to compensate the dominant harmonic of the line current. The system automatically compensates the reactive and dominant harmonic power flow in order to guarantee the desired displacement power factor (DPF) and a low harmonic distortion. No information about power system parameters is necessary to the correct operation of the topology. Single phase practical results are presented and discussed.