ABSTRACT
This chapter considers the use of neural network (NN) to improve the performance of a distributed generation (DG) system with active power filtering (APF) capability. The correct DG connection to the grid depends strongly on the correct detection of the fundamental harmonic frequency and phase of the voltage at the coupling point. The chapter discusses the results obtained experimentally on a 130 V RMS, 50 Hz single-phase power distribution network, which in itself contains the distorted currents arising from both the utility and a nonlinear load. The performed tests consist in the insertion test of APF with the nonlinear load and with the auxiliary nonlinear load using firstly the p-q theory–based algorithm and secondly the NN adaptive filter–based control. The nonlinear load current is obtained with a diode bridge supplying a highly inductive load, so the required current is almost a square wave.
