摘要

There are abundant information after power system signals are processed by wavelet transform, so it is necessary to research appropriate method to perform post-analysis of a great deal of information from wavelet transform and extract their features. From the point view of quantitative analysis, the authors present the concept of the post-analysis of wavelet transform, describe several feature extraction methods for power system transient signals after wavelet transform, i.e., the post-processing method including module maxima and singularity analysis, energy distribution analysis, wavelet coefficient clustering analysis, wavelet coefficient statistic analysis, and wavelet entropy analysis, and the physical meaning of applying them to power system fault detection and classification. The application of a new kind of post-analysis method, namely wavelet singular spectrum entropy, in frequency-varying system is simulated. Simulation results show that this post-analysis method can be applied to fault detection of power transmission lines and equipments.

全文