An application of neural network in distribution system harmonic detection

In this paper, an effective tool is proposed to detect harmonic components by using probabilistic neural network (PNN). PNN is used to detect the harmonics from the distorted waveforms. PNN can be fast learning and recalling process, no iteration for weight regulations in the learning process, no pre-decision for the number of hidden layers and the number of hidden nodes in each layer, and adaptability for architecture changes. Many tests are conducted and the results show that PNN has advantages over other previously developed algorithm. It provides a simplifying model and shorten processing time to detect harmonics.