Optimized Control of DC Voltage Based on BP Neural Network in Active Power Filter

Abstract

This paper presents an intelligent method of optimized control of DC voltage based on BP Neural network, on account of this presentd method, by collecting parameters of grid voltage and significant frequency of harmonic current, then, construct couples of tranining samples and fully use the advantages of BP Neural network that can discretionarily utlize arbitrary precision to approximate any continuous function, and computing the optimized voltage of DC side, as a result, achieving online adjustment of capcitors voltage at DC side. Futher more, adjusting the hysteresis width automatically according to the voltage variation of capacitor, thus the voltage of DC side can automatically adapt to fluctuation of grid voltage and load current, so that to obtain comprehensive optimum of power loss and compensation performance of power active filter.