Harmonic Detection Using the Direct Weight Determination Neural Network

Abstract

This paper presents a novel harmonic detection algorithm using the direct weight determination neural network for the electric power system. A new ANN structure is designed to strengthen the real-time capability of harmonic detection. The proposed algorithm employs the weight computation with sine base function to address the problem of harmonic detection. The optimal weight of this ANN with sine base function can be achieved by direct computation. This ANN can avoid the tediously long weight training and get the proper weight including the information of phase and amplitude of harmonic detection. The simulation computation demonstrates this algorithm has high precision and low computational complexity, and it has value in the harmonic detection of electric power system.