Probabilistic Neural Network Based Reference Current Generation for MLI Based SAPF in Harmonics Reduction

Due to a sensitive irregular load integration, the demand for better Power Quality (PQ) in distributed power systems has increased. To enhance the system stability by reducing the harmonics, Active Power Filters is widely utilized. A usage of Shunt Active Power Filter (SAPF), which removes load current harmonics and supplies load reactive power to provide harmonic free source currents with a power component, is suggested in this paper. AC source output is fed into Decoupled Double Synchronous Reference Frame (DDSRF) which generates sinusoidal harmonics in opposition to the load harmonic. The influence of harmonics is then lowered by injecting this harmonic into the power systems. Probabilistic Neural Network (PNN) classifier generates a harmonics free reference current signal. The actual and reference voltage inputs are given to Proportional Integral (PI) controller which minimizes signal error and an output is fed into PWM generator to produce gate pulse for the operation of 5-level Cascaded Multilevel inverter. The currents from AC source and opposite current obtained from the MLI is injected into PCC and fed into nonlinear load. The proposed paper is executed in MATLAB simulation software and corresponding results are obtained particularly.