Power quality disturbance signals identification based on wavelet packet and SVM

Identification and classification of fault signal in power systems is an important task. For the pattern recognition of disturbance signal, this paper presents five benchmarks of disturbance signal by MATLAB; Then, feature vectors of the disturbance signal which are extracted by the wavelet packet transforms; The vectors can be recognize by SVM multi - classifier. Numerical results show this approach achieved a classification accuracy of 97%, a few training samples and training time is short, a good real-time performance. It is an effective method for identify power disturbance signals.