Classification for voltage sags based on similarity of generalized S-transform modulus matrixes

This paper proposes a new classification method about voltage sags based on the module matrix of generalized S-transform. It firstly determines the adjustment factor λ of Gauss window width according to the main frequency constituents of the signals, then calculates the modulus matrixes of the S-transform so as to form the standard templates. The signals can be classified by the analysis of the similarity between the specific region of the matrixes of the test signals and the standard templates. In the process of similarity comparison, for enlarging the differences among templates, the concept of normalization and the idea of calculating similarities at different frequencies are proposed. This method sufficiently excavates the different disturbance characteristic differences; no extra classification processes are needed. Simulation analysis indicates that this method is easy to be implemented and has high accuracy and noise robustness.