Research On Faults Diagnosis Method Of Spring Operating Mechanism Circuit Breakers Based On Signals' Multiple Dimension Feature Extraction

High voltage circuit breakers(CB)are one of the most important power equipment in power system,the reliable action of which is significant for the safe and stable operation of power grid.Therefore,it is crucially needed to takes measures to realize real-time condition monitoring of them.Based on the field investigation and literature review,this paper determined several typical high-incidence faults during CB's operation,which concludes faults in trip and close coil circuit,faults in operating mechanism,and faults in transmission mechanism.The first kind of faults and the post faults are simulated respectively on two different types of CB equipment due to limitation of objective conditions.For the faults occurring in CBs' trip and close coil circuit,this paper performs large quantities of fault simulation tests to acquire sufficient number of coil current signals and dynamic contact displacement signals,after which the smoothing filter and median filter are applied in signal denoising and effective current,time,displacement,and speed characteristic value are extracted from the processed signal to serve as the object of study.First,multidimensional characteristic is used to build the IF-THEN classification model to test the known samples;Then,multidimensional characteristic is used to train and test the support vector machine,selecting the C-SVM based on radial kernel function as the target SVM model;Finally,by comparison,IF-THEN classification model is not good at distinguishing between similar faults while SVM has a high accuracy in the faults diagnosis.Operating mechanism provides power for CB's action,while transmission mechanism transmits power to the action element.The typical structure of the two mechanism is the energy storage spring and cam,the former apt to fatigue during long-time compression or frequent tension-compression while the latter prone to wear and tear after CB's action times accumulated to a certain amount.In this paper,mass faults simulation tests were conducted and sufficient numbers of vibration signals were collected,including three mutually perpendicular directions in space,after which wavelet packet decomposition and reconstruction are applied to the vibration signals and the wavelet packet energy in each frequency band is obtained and serve as the object of study.First,the wavelet packet energy of all bands is used to train and test SVM;Then,wavelet packet energy is screened by K-S test method to realize reduction of samples' dimension,and the new formed samples are used to train and test SVM.Finally,by comparison,the wavelet package energy screened by K-S test method has a better separability,and can more accurately reflect the two kinds of typical mechanical faults in CB.