Identifying the Overvoltage in Distribution Networks Based on Support Vector Machine

摘要

The internal and external overvoltage in distribution networks are the main reasons of accidents. Based on overvoltage data in distribution network recorded by overvoltage on-line monitoring system, this paper analyzes the characteristics of zero sequence voltage waveform. The maximum amplitude and RMS value of zero sequence voltage, the minimum RMS of low frequency component of three phase voltage in stable range and other five parameters are selected as characteristic parameters to identify internal and external overvoltage. According to the operation data and record, all overvoltage data record by on-line monitoring system were labeled as internal or external overvoltage and the discrimination function is constructed by support vector machine method. The test results of field acquired overvoltage data indicate that the identification parameters and method base on the zero sequence voltage and SVM are correct and effective.