Circuit breaker opening/closing springs have different degrees of fatigue for a long time running

duringthe power system failure

it is possible to have a major accident if not cut off the line and the failure timely. In orderto detect the types of the fault degree of high voltage circuit breaker

quantify the fault degree of the circuit breakerspring and gives the measures of whether circuit breaker can continue to use. This paper analyzes the circuit breakertravel signal in mechanical vibration signal and proposes a method to process the travel signal. Processing the travelsignal to obtain velocity and the variance of volatility distribution and form a feature vector P=[v

σ]. Using the serialdecision to support vector machine（support vector machine

SVM）to classify the degree of fault of spring and de-fine the fitting function with variance of the spring as the independent variable and the variance of the distribution toexpress the spring fault diagnosis concretely

it turns out this method can diagnose high-voltage circuit breakerspring failure effectively. And the fitting function based on wave variance can judge the state of the spring fully. It isof practical significance to prevent circuit breaker failure.