Analysis of power grid harmonics with wavelet network

Harmonics are usually analyzed through Fourier transform or wavelet transform, both of which have their shortcomings. Wavelet neural-network is a new algorithm that combines wavelet analysis with neural-network, suitable for processing signals with time-varying frequency domain characters. To analyze the third and fifth harmonics, a wavelet network structure was constructed, training algorithm described in detail, simulation experiments showed very satisfactory results, proving the algorithm to be good at analyzing harmonics in real-time with high accuracy.