Harmonic estimation using radial basis function neural network

Abstract

Harmonic estimation is the initial and main process for designing filter for effective harmonic
mitigation. A democratic neural network technique of radial basis function neural network (RBFN) is
proposed to estimation of harmonics and the outcomes are compared with the established ofBack
Propagation Neural Network (BPN). The BPN has the demerits of slow convergence, repeated training
required to reduce the error which results in time consuming and also not meeting the performance goal
with complex systems with more number of neurons. The simulation results show that the proposed RBFN
method is accurately predicting the harmonics at PCC with very less computational time and error when
compared with the BPN network.