An application of radial basis function neural network for harmonics detection

The increasing use of nonlinear loads such as power electronic devices has led to serious harmonic pollution in the power system. In order to prevent harmonics from deteriorating the power quality, detecting harmonic components for harmonic mitigations becomes an important issue. In this paper, the radial basis function neural network (RBFNN) suitable for function approximations and pattern classifications is used to identify harmonics. Simulation results are compared with those obtained by using the fast Fourier transform (FFT) and the back-propagation network (BPN). It is shown that the proposed solution procedure yields relatively more accurate results, while the computational efficiency is maintained.