Research On Non-invasive Diagnosis Technology For Mechanical Fault Of Conventional Circuit Breakers

Conventional circuit breaker is one of the key equipments for low voltage distribution network in power system,widely used in the occasions of the total input-wire and important equipment for distributing system of low voltage.The practice shows that circuit breaker occurs mechanical problems easily.The existing research work mainly focused on the mechanical fault diagnosis of high voltage circuit breaker,but the operation reliability of conventional circuit breaker with the characteristics of low voltage and large current is also related to the safe operation of power system.Therefore,the research on the non-invasive mechanical fault diagnosis technology of conventional circuit breaker has the vital significance.The non-invasive experimental system of conventional circuit breaker is constructed to measure mechanical states through artificial simulated failure of circuit breaker.According to the characteristics of vibration signal,sound signal and current signal of energy storage in the process of switching,different fault diagnosis methods during action process are given in order to realize the non-invasive diagnosis of circuit breaker mechanical failure.The main research contents and results are as follows:1.A kind of non-invasive experimental system for conventional circuit breaker based on LabVIEW is constructed.The fault simulation experiments are carried out on the conventional circuit breaker of DW15-1600 to realize the control of the conventional circuit breaker action,the acquisition of vibration signal,sound signal and current signal,data storage,signal preprocessing,and other functions.And it has good interoperability and scalability.2.A method that combines complementary ensemble empirical mode decomposition(CEEMD)-sample entropy and relevance vector machine(RVM)is proposed for fault diagnosis by using vibration signal during the switching process.By using the vibration signal containing abundant information of mechanical properties as a signal source,sample entropy was extracted from the intrinsic mode functions to form feature samples.Euclidean distance of different fault samples is calculated to establish the binary tree classifier based on RVM.The method realizes the accurate identification of switching fault type with relatively less fault data samples.3.In order to guarantee the reliability and stability of the fault diagnosis under harsh conditions,on the basis of fault diagnosis based on vibration signal,a vibration and acoustic joint diagnosis method on the switching fault of conventional circuit breakers based on multi-feature fusion and improved quantum particle swarm optimization(QPSO)-relevance vector machine(RVM)is proposed.Using the complementary of vibration signal and acoustic signal,multi feature parameters are extracted to solve the problem of low recognition accuracy and low stability of single feature.And the improved QPSO was used to optimize classification model parameters to improve the accuracy of the binary tree model based on RVM.4.A fault diagnosis method of conventional circuit breaker operating mechanism based on current signal analysis of the energy storage motor is proposed.By monitoring current signal of the energy storage motor,Hilbert amplitude demodulation method and improved threshold wavelet packet method are used to obtain the envelope of the AC current signal,which can solve the rough problem of the extracted envelope caused by random noise interference.The fuzzy clustering and relevance vector machine(RVM)are combined to realize the non-invasive fault diagnosis method for operating mechanism of conventional circuit breaker.