Mechanical Fault Diagnosis Of High-Voltage Circuit Breaker Based On Vibration Signal Processing

The high-voltage circuit breaker plays a role of control and protection in the power system.Its safe and reliable working state is related to the safe and stable operation of the power grid.According to the investigation and research of the relevant organizations at home and abroad,it is found that the power failure loss caused by the high-voltage circuit breaker fault far exceeds the value of the equipment damage.When the operating mechanism of high-voltage circuit breaker acts,the released mechanical vibration signal contains abundant equipment status information.Therefore,the mechanical fault diagnosis of high-voltage circuit breaker can be realized by analyzing the vibration signal.In this paper,the 12 kV VD4 intelligent high-voltage circuit breaker of ABB company is taken as the research object,based on the vibration signal monitoring means,the vibration signal acquisition system of high-voltage circuit breaker is established.Aiming at the four most common mechanical faults of VD4 circuit breaker,four parts of research contents are put forward:acquisition system construction and signal acquisition,signal pre-processing,signal feature extraction and circuit breaker state identification.Firstly,four common mechanical faults of VD4 intelligent high-voltage circuit breaker are counted:mechanism jam,linear spring fatigue,lubrication failure and buffer fault.According to the vibration frequency of mechanical fault,the appropriate acceleration sensor is selected and the best installation position is determined.The vibration signal acquisition system is built,and the collected vibration signal is denoised by wavelet threshold.Secondly,two methods are introduced to extract the features of signals:(1)Ensemble Empirical Mode Decomposition-Approximate Entropy(EEMD-ApEn)feature extraction method,the signals are decomposed into IMF components of different layers,and the IMF components strongly related to the original vibration signals are selected for ApEn calculation,and the vectors formed by these ApEn values are the features;(2)Wavelet Transform-Local Binary Patterns(WT-LBP),which will transform the signals into gray-scale images in each region,the frequency of the LBP value is counted,and the histogram normalization is the feature vector.Finally,two kinds of characteristic signals are used for fault identification,and support vector machine(SVM)classification and BP neural network classification are used.Through the comprehensive analysis of detection rate and false detection rate,we get the best diagnosis method:Based on EEMD-ApEn feature extraction and SVM classification,the detection rate is 99.5%,false detection rate is 0%,the diagnosis effect is good.The experimental results show that the fault identification method proposed in this paper based on vibration signal provides a new scheme for the health management of high-voltage circuit breakers,further realizes the accurate decision-making,and embodies the high application value.