Vibration Signal Analysis And Fault Recognition Of Low-voltage Breaker

Low-voltage circuit breaker (LVCB) is an important equipment of low voltage distribution networks, which play a role of control and protection. Its’an important guarantee of low voltage distribution networks that the stable running state of LVCB. Therefore, the monitoring of the running states of LVCB and the diagnosis of fault trend is especially important to the safety of the power grids. This paper makes a research on the mechanical status of LVCB to detect the running status and prognosis fault, since mechanical fault is the main fault of breaker. The vibration signal acquired during LVCB operation contains a large number of device state information. It can extract useful information of the circuit breaker state by analyzing the vibration signal. This paper takes use of modern information processing technology to analyze the LVCB’s vibration signal and extracts the useful mechanical state features as a judgment basis of fault diagnosis.Firstly, in order to obtain vibration signal of LVCB, this paper built a data acquisition circuit, which can acquire the state information such as vibration signal, current of energy-stored motor, contactors’signal under different conditions. Combined with it, the LVCB data acquisition system is developed by the Lab VIEW software.Secondly, on the problem of vibration signal feature extraction, this paper proposed a method which combines the empirical mode decomposition with fractal dimension for extracting the feature of vibration signal based on summarizing existing vibration signal processing technology. The empirical mode decomposition is used to decompose vibration signal in different frequency bands, and the fractal dimension is an important state characteristics of the nonlinear system. Each component’s fractal dimension of empirical mode decomposition will be changed when LVCB fault occurred. So the component’s fractal dimension of empirical mode decomposition can represent state characteristics of LVCB. And this method is used to analyze the vibration signal feature of switching synchronization of LVCB.Finally, this paper adopted neural network to diagnose the LVCB fault. Taking the component’s fractal dimension of empirical mode decomposition as fault feature for the training and testing of neural network. On the basis of comparing the BP neural network, extreme learning machine as the diagnostic network of LVCB fault is introduced. The extreme learning machine is a new type of neural network which used the analytic method to calculate the output weights of network all at once. Extreme learning machine overcomes the BP neural network’s disadvantage such as slow training speed and easily trapped in the local optimum. In addition, the extreme learning machine is applied to identify the different faults, and achieved good results.The low-voltage circuit breaker data acquisition system based on Lab VIEW can also be applied to acquire the parameters of other electrical equipment. Moreover, further research on the method in this paper can be used in online monitoring of LVCB, and it has a good application prospect for the LVCB status maintenance.