Vibration Signal Processing And Fault Diagnosis Of 12kV Intelligent Vacuum Circuit Breaker

The vibration signal feature knowledge base of 12 kV intelligent vacuum circuit breaker was established after adopting several vibration signal processing methods in this paper.Also,the fault diagnosis model of intelligent vacuum circuit breaker was established based on the algorithm of support vector machine(SVM)and vibration signal feature knowledge base.In search of proper indicator of the some specific faults from the opening stroke curves and closing currents,the fault simulation tests of the damper and closing release were designed.(1)The mounting position and mounting method of the piezoresistive acceleration transducer were designed as well as the amplifying circuit of the vibration signal.The contrastive analysis of waveform,obtained by piezoresistive transducer and piezoelectric transducer respectively,showed that the performance of the piezoresistive transducer is better with faster response rate and higher sensitivity,the collected vibration signal waveforms are more fine and the vibration process is more distinct.(2)Several denoise methods of the vibration signal,including difference spectrum theory of signal value,wavelet-packet energy method as well as wavelet thresholding de-noising method,were adopted.The contrastive analyses of denoise methods showed that wavelet thresholding de-noising method could not only decrease the noise,but also retain the information of the original signal as much as possible.(3)Several kinds of vibration signal energy entropy were calculated by different methods,including wavelet packet decomposition,ensemble empiricalmode decomposition(EEMD)and short time Fourier transform(STFT).After comparing vibration signal energy entropy above,the method of subsection statistics of signal peak value,which took the parameters of the generalized extreme value(GEV)distribution function as vibration signal characteristic parameter,was put forward to processing vibration signal.The result of processing showed that it could well distinguish the state of the circuit breaker by utilizing the method of subsection statistics of signal peak value,which could be used to extract characteristic parameter of the vibration signal.(4)The data set of vibration signal characteristic parameter was established after selecting its characteristic parameter.The result of the test and performance analysis of SVM fault diagnosis model which established by samples of the knowledge base,showed that SVM model with better performance of learning was able to realize the fault diagnosis of the intelligent vacuum circuit breaker.Also,the accuracy of the SVM fault diagnosis model could be improved by selecting and normalizing sample of the train set.(5)The fault simulation tests of the damper and closing release were conducted.The corresponding characteristic parameters,contrasted from opening stroke curves and closing current under different states,had been fitted with function.It arrived a conclusion that the characteristic parameters above will be the appropriate indicator for recognizing the fault of the damper and closing release.