On-line harmonic estimation in power system based on sequential training radial basis function neural network

Harmonic estimation is considered the most crucial part in harmonic mitigation process in power system. Artificial intelligent based on pattern recognition techniques is considered one of dependable methods that can effectively realize highly nonlinear functions. In this paper, a radial basis function neural network (RBFNN) is used to dynamically identify and estimate the fundamental, fifth harmonic, and seventh harmonic components in converter waveforms. The fast training algorithm and the small size of the resulted networks, without hindering the performance criteria, prove effectiveness of the proposed method.