POWER QUALITY IMPROVEMENT IN DISTRIBUTION SYSTEM USING ARTIFICIAL NEURAL NETWORK BASED HYBRID POWER FILTER

Abstract

Nowadays, power quality has become an important concern in industrial distribution power system operation due to the presence of increased power electronic converters with nonlinear loads and linear loads. The nature of nonlinear loads generates harmonics in the current waveforms whereas the linear loads also causing lagging power factor. The harmonic current and low power factor problems exist in the industry results from the nonlinear and inductive linear loads, which need critical care for their compensation in the ONTEX Ethiopia PLC factory. The proposed hybrid power filter is composed of a shunt active power filter and a shunt passive power filter. The hybrid power filter is connected in parallel with the power line and is used to mitigate the current harmonics and compensate reactive power to improve power factor. The artificial neural network and PI current controller are applied to control the shunt active power filter and SPWM to generate switching signals to turn ON and OFF the IGBT as fast as possible. The ANN current controller is trained, tested, and validated using the Levenberg Marquardt backpropagation algorithm to control the current of the voltage source converter. The MATLAB Simulink has been used to demonstrate the overall system for harmonic mitigation, reactive power and power factor correction. The result of simulation shows that the total harmonic distortion (THD) current reduced from 27.51% to 4.39% using hybrid power filter based PI current controller and from 27.51% to 2.98% using hybrid power filter based ANN current controller. The application of ANN based current controller improved the power factor from 0.72 to 0.964. In addition, the application of DC link voltage regulation significantly enhanced the overall power quality