Research On Mechanical Fault Diagnosis Method Of Circuit Breaker Based On Vibration Signal

The normal working state of the circuit breaker is the basis of safe and stable operation of the power grid.If the circuit breaker works in the fault state,it will have a serious impact on the normal operation of the power grid.The mechanical structure of the circuit breaker generates a vibration signal containing rich information during the operation,and the mechanical failure is the main failure type of the circuit breaker.Therefore,the vibration signal can be analyzed to achieve the purpose of mechanical fault diagnosis.In order to measure the vibration signal of the circuit breaker,an experimental platform for the vibration signal detection system was built.The iron core sticking fault,the buffer invalid overpass fault and the insulating tie rod loosening fault are respectively set by the method of artificially setting the fault.In the normal state and three different fault conditions,the vibration signal generated by the circuit breaker during the operation is measured.In order to accurately extract the vibration signal from the environment that is submerged by noise,a signal denoising method based on sparse decomposition is proposed.The fast fourier transform-matching pursuit algorithm is used to denoise the signal.And the vibration signal of the four states of the circuit breaker is used to compare and analyze the denoising performance of the fast fourier transform-matching pursuit algorithm and the wavelet denoising method.It is proved that the fast fourier transform-matching pursuit algorithm has certain advantages in the denoising of the vibration signal of the circuit breaker.Different kinds of feature extraction methods correspond to different types of fault features.If only a single type of method is used for feature extraction,the extracted features will lack specificity and it is easy to ignore certain fault features.So that the type of fault diagnosed is not complete.In order to increase the specificity of features.a new time-frequency-data sequence feature extraction method is proposed.The time-frequency method combining singular value decomposition and total least squares-estimation of signal parameters via rotational invariance techniques is used to extract time-frequency features with clear physical meaning.The data sequence method combining variational mode decomposition and multifractal is used to extract data sequence features without clear physical meaning.The time-frequency features and data sequence features are combined to obtain new time-frequency-data sequence features.The performance of time-frequency feature extraction method,data sequence feature extraction method and time-frequency-data sequence feature extraction method is compared by using the vibration signals in four states of the circuit breaker.It is proved that the time-frequency-data sequence feature extraction method has certain advantages in dealing with circuit breaker fault diagnosis.The classification performance of the support vector machine algorithm and the random forest algorithm is compared by using the vibration signals in four states of the circuit breaker.Considering the factors such as fault recognition rate and ease of operation,the random forest algorithm has better classification performance and better applicability in circuit breaker fault diagnosis.