The improved training algorithm for prediction of short-circuit based on signal processing method

With increasing needs of utility industry, the improved technology leads to lower cost per power both for small and large units. The power quality disturbances and security constraints for the power network are important issues of the transmission and distribution system construction. A novel approach using signal processing method is introduced for prediction of short-circuit fault. The wavelet transform provides a framework which can locate energy in time-frequency domain for feature extraction of the disturbance signal, where multiresolution analysis is an excellent tool in providing spatial-frequency decomposition. The application of signal denoising based on the statistic rule is brought forward to compute the threshold of each order of wavelet space, determining the decomposition level adaptively. The effectiveness of the proposed method to extract the characteristic signal is described, which can be realized by the calculation process of those types of short-circuit fault. The simulated results obtained support the effectiveness of the proposed signal processing method.