AR Prediction Method of Unbalance Factor Index of Power Quality Based on Discrete Fourier Decomposition

An autoregressive (AR) method based on discrete Fourier analysis is proposed to forecast unbalance factor index of power quality. The discrete Fourier transform of the sequence of unbalance factor index is analyzed, and the low frequency components are extracted. An autoregressive method is used to forecast each low frequency component of index sequence. Through inverse discrete Fourier transform, the forecasting low frequency components are converted to forecasting sequence of the unbalance factor index in time domain. The proposed method is verified by actual data, and the results show that discrete Fourier analysis can reduce the influence of high frequency noise. By comparing with BP neural network prediction method, the AR method based on Fourier transform can more effectively forecast unbalance factor index.