ABSTRACT
A surge arrester fault diagnosis method based on Support Vector Regression (SVR) is put forward. In the method, the influence of environmental factor and power network factor on leakage current value is taken into account and as the input feature vector of support vector machine, and the fundamental and third harmonic waves of resistive current are taken as the diagnosis characteristic of surge arrester dampness and aging respectively. The regression model of unaged surge arrester is determined via multivariable experiment, and the predictive value of the diagnosis characteristic is obtained in combination with online monitoring data at the time of diagnosis. With the predictive value and the measured value, the deviation between the current state and the unaged state can be calculated, and then the occurrence and severity of fault can be judged on this basis. At last, the method is verified with an example for accuracy and effectiveness.
