Neural Network-Based Selective Compensation of Current Quality Problems in Distribution System

Active power filters (APFs) have been used to compensate harmonics, reactive current, and negative sequence fundamental frequency current drawn by nonlinear loads. The control of APF is the core issue for their proper operation. The flexibility of selective compensation embedded in the control scheme makes APF versatile for compensation of reactive power, harmonic currents, and unbalance in source currents and their combinations, depending upon the limited rating of voltage source inverter employed as APF. The proposed scheme utilizes neural network-based decomposition of the load current into positive and negative sequence fundamental frequency component, reactive component and harmonic components. The adaline-based current decomposer estimates the reference currents through tracking of unit vectors together with tuning of the weights. The implementation of the control scheme facilitates selective compensation which respects the limited rating of the APF. The simulated results using developed MATLAB model are presented and are validated by experimental results to depict the effectiveness of the proposed control method of APF