Reference signal generator for active power filters using MGP-FIR filter designed by evolutionary programming

Abstract

This paper describes a high-performance reference signal generator for active power filters extracting the fundamental signal component from distorted current signals. In order to achieve high-quality output as well as computationally effective algorithm, the generator employs an adaptive and predictive MGP-FIR (Multiplicative General Parameter) bandpass filter designed by evolutionary programming. Detailed procedures of MGP-FIR filtering and evolutionary optimization are first discussed; theoretical conclusions are verified by illustrative simulation results.