Power quality disturbance identification based on clustering-modified S-transform and direct support vector machine

A method based on CMST(Clustering-Modified S-Transform) and DSVM(Direct Support Vector Machine) is proposed to identify the power quality disturbance. Combined with the characteristics of power quality disturbance signal, the CMST method can optimally and simultaneously process the time-domain resolution of fundamental frequency signal and frequency-domain resolution of high-frequency signal to ensure the correctness of property extraction. Compared with the least squares support vector machine, DSVM, as a classifier, has simpler solving process, lower computation complexity, faster training and testing speed, higher generalization ability. Furthermore, it guarantees the global optimal solution. The CMST combined with DSVM is applied in the identification of single or mixed disturbance. Simulative experiment verifies the effectiveness of the proposed method. ©, 2015, Electric Power Automation Equipment Press. All right reserved.