Research On Vibration-Signal-Based Monitoring Method Of Mechanical State In Spring-Operation-Type High Voltage Vacuum Circuit-breaker

The vibration signal in circuit-breakers contains abundant mechanical state characteristics.Therefore,after obtaining vibration signal through advanced sensing technology,we are able to recognize the mechanical state inside the circuit-breaker by using signal analyzing technology and state recognition algorithm.Due to its non-invasive acquisition characteristic,the vibration signal is high voltage isolable and has a small attenuation in metal devices,which makes the vibration-signal-based mechanical state monitoring in circuit-breaker attract great attentions and intensive investigations from worldwide researchers.On the basis of experimental investigation,the vibration-signal-based monitoring method of mechanical status in spring-operation-type high voltage vacuum circuit-breaker,as well as the state recognition algorithm are thoroughly studied in this paper.In this paper,a spring-operation-type high voltage vacuum circuit-breaker ZW10-12 / 630 A is chosen as the research object.By building an experimental platform for mechanical state monitoring,the optimal installation position of acceleration sensor and the hardware selection for state acquisition are well studied.Besides,the computer analysis system software for the experimental platform is designed by LabVIEW virtual instrument.The experimental platform is expected to lay a solid foundation for acquisition and noise elimination of vibration signal,as well as extraction and identification of mechanical state characteristics.Generally,the various noises in the vibration signal acquisition site will make the gathered vibration waveform contain a large number of high-frequency white noise.Considering this,the wavelet de-noising method is studied in this paper,with the focus on analyzing soft threshold selection method suitable for circuit breaker vibration signal de-noising.By comparing the wavelet de-noising effects of four soft threshold selection rules with the assistant of MATLAB,it is found that the wavelet de-noising method based on the mixed threshold selection rule is suitable for the circuit breakers vibration signal de-noising.Besides,by using the wavelet packet transform method to analyze the de-noised vibration signal,which includes analyzing the high/low frequency characteristics of vibration signal under different scales and reconstructing vibration signal in each node and each layer,the weight of vibration signal in different waveband can be obtained.Furthermore,since the energy value of vibration signal in each frequency is able to characterize the different mechanical states of circuit-breaker,by using waveband analyzing technology to extract the energy value of vibration signal in each frequency,the extraction of mechanical state characteristic is then completed.After analyzing the defects of BP neural network,a quantum genetic algorithm is proposed to seek the optimal initial value of BP neural network,and the obtained quantum genetic BP neural algorithm based state recognition algorithm is well studied.By using function maxim optimization in MATLAB,the comparison of optimizing capacity between quantum genetic algorithm and genetic algorithm is realized.Moreover,the convergence effects of BP neural network and quantum genetic BP neural network are compared through 3-7-3 decoder test experiment in MATLAB,verifying the feasibility of improved state identification algorithm.Applying quantum genetic BP neural network to the vibration-signal-based monitoring of mechanical state in spring-operation-type high voltage vacuum circuit-breaker,a conclusion can be arrived,according to the simulation results,that the improved state identification algorithm owns more accurate evaluation results,less iteration error and better convergence effect than the conventional BP neural network.