Research on UPFC fault diagnosis based on wavelet transform and support vector machines

At present, the way of dealing with UPFC faults in the power grid is accomplished by relay protection. As the protection equipment removes the entire UPFC system when the UPFC fails, the maintenance personnel cannot get the fault information. In response to the problem of long fault time and low reliability, a fault diagnosis method is proposed in this paper. After processed for twice, the UPFC DC side voltage and current signals are selected to be analyzed by using db2 wavelet. The fault points are found by the 6th layer wavelet signal and then the eigenvalues of the electrical signal waveforms are extracted. The eigenvectors consisted of eigenvalues are classified by the support vector machine (SVM) to realize the fault diagnosis of UPFC. The results of UPFC fault diagnosis provide maintenance basis for maintenance personnel, so that maintenance work is more targeted and the recovery is faster.

I. Introduction

With the development of social economy, the load center gradually becomes away from the energy center, therefore, long-distance, large-capacity transmission becomes the inevitable development of the power grid. It is limited of traditional power system operation adjustment means to flexibly, effectively and timely control power system. So flexible control of the FACTS equipment has become an effective way to solve the problem of modern power system. As a third generation FACTS device, UPFC can achieve parallel compensation, series compensation, phase shifting, terminal voltage adjustment and other basic functions which can accomplish real-time control and dynamic compensation for the AC transmission system^[1]–[3]. Therefore, UPFC is the key technology to solve the current power grid problem.