Neural Network and Bandless Hysteresis Approach to Control Switched Capacitor Active Power Filter for Reduction of Harmonics

This paper proposes a combination of neural network and a bandless hysteresis controller, for a switched capacitor active power filter (SCAPF), to improve line power factor and to reduce line current harmonics. The proposed active power filter controller forces the supply current to be sinusoidal, in phase with line voltage, and has low current harmonics. Two main controls are proposed for it: neural network detection of harmonics and bandless digital hysteresis switching algorithm. A mathematical algorithm and a suitable learning rate determine the filter's optimal operation. A digital signal controller (TMS320F2812) verifies the proposed SCAPF, implementing the neural network and bandless hysteresis algorithms. A laboratory SCAPF system is built to test its feasibility. Simulation and experimental results are provided to verify performance of the proposed SCAPF system.