Research On Fault Diagnosis Of Miniature Circuit Breaker Based On Fractal Theory And Probabilistic Neural Network

As a terminal distribution electrical appliance,miniature circuit breaker is not only an important guarantee for personal and property safety,but also a key component of the power distribution system.Its good operating state is a necessary prerequisite for ensuring the safety of users.Therefore,the research on fault diagnosis of miniature circuit breakers has very important research significance and value.The faults of miniature circuit breakers are mostly mechanical faults.The vibration signals during the closing and opening of the circuit breakers contain their state characteristic information.The vibration signals of different faults will have certain differences so that it can analyze and process the status information of the circuit breaker.This paper mainly analyzes and studies the mechanical state of miniature circuit breakers to realize the fault diagnosis of circuit breakers.It also analyzes and processes the vibration signals of low voltage circuit breakers collected by modern information processing technology,and extracts the effective mechanical state feature quantity as the fault category basis for judgment.The empirical mode decomposition method is analyzed,and the local features of the signal are decomposed into several Intrinsic Mode Function components according to the frequency.The empirical mode decomposition effect is proved better than the wavelet decomposition.The vibration signal of the miniature circuit breaker is processed by this method.The G_P algorithm is used to study the method of obtaining the correlation dimension in fractal dimension.The selection method of delay time and embedding dimension in G_P algorithm is analyzed.The methodof combining empirical mode decomposition and fractal theory is used to extract the characteristic value of the vibration signal of miniature circuit breaker.The accelerometer is used to collect the three-phase fault vibration signal of the miniature circuit breaker under the experimental simulation.The neural network is used to train and test the fractal dimension of each component after the empirical mode decomposition as the fault feature vector.On the basis of comparing BP neural network,probabilistic neural network is introduced as the diagnosis network of miniature circuit breaker faults,and Bayesian optimal results are obtained,and the classification is pretty accurate.Finally,the vibration signals of the small circuit breakers under different faults are decomposed according to the method studied in this paper,and the fractal dimension is extracted.The probabilistic neural network is used for fault diagnosis analysis.Experiments show that the method can achieve the expected goal,and further study of the condition monitoring can be applied to miniature circuit breakers,which has a good application prospect.