Condition Monitoring And Diagnostic Of High Voltage Circuit Breaker Based On Mathematical Morphology And Wavelet Decomposition

With the rapid increase of power loads in modern power systems,the use and maintenance of high-voltage circuit breakers are receiving more and more attention from operation and maintenance personnel.The mechanical failure,such as coil aging,core catching,transmission jamming and cam wearing,affects the security and stability operation of the power system,and causes huge economic losses.However,there does not exist a valid and reliable system that can diagnose the faults timely and accurately.Hence,it is of great importance to monitor the mechanical condition of the circuit breaker,recognize the monitored data accurately and analyze the operation status,fault nature and location.This paper proposes the concepts of multi-parameter monitoring and multi-dimensional mapping diagnostic,in order to perform extensive current condition monitoring,effective feature extraction,and complete fault diagnosis.This thesis firstly introduces the objectives of mechanical condition monitoring of circuit breakers,including opening and closing coil current,energy storage motor current and vibration signal.The third chapter describes online monitoring methods and data acquisition.A laboratory test platform based on LabVIEW FPGA is built to obtain circuit breaker mechanical failure data by fault simulation.The fourth chapter describes in detail feature extraction of the monitored data using mathematical morphology and wavelet packets,and the choosing of feature information.Finally,the principal components of the feature information extracted from the data obtained by fault simulation are used to build a multi-mapping knowledge base.Support vector machine(SVM)is used to diagnose the mechanical status of the circuit breaker.In this process,the impacts of a failure on the multiple parameters are comprehensively considered.This thesis proposes a new signal processing method of vibration signal in particular.Wavelet transform is combined with relative energy and entropy to extract the feature information of a vibration signal,which in cooperation with the feature information of currents is employed to diagnose the mechanical state of the circuit breaker.Through comprehensive laboratory experiments,it can be concluded that the methods proposed in this thesis can effectively monitor and diagnose the common mechanical problems of high-voltage circuit breakers.