基于IEMD和GA-WNN的断路器分合闸线圈故障诊断方法

Fault Diagnosis Method for Circuit Breaker Opening and Closing Coil Based on IEMD and GA-WNN

真空断路器二次回路或操动机构运行状态能通过电流曲线特征反映。首先,通过对真空断路器分合闸线圈铁心卡涩、电压异常(过高或过低)和击穿3种常见故障进行实验室模拟,创建了故障电流曲线特征库。其次,利用故障电流信号经过经验模态分解后的经验模态分量中的能量密度乘对应平均周期为恒定常数的性质,提出一种改进经验模态分解方法来提取分合闸线圈电流特征值,并将其作为小波神经网络的输入样本集。并在此基础上,提出一种改进遗传算法与小波神经网络结合的断路器故障诊断方法。该方法利用改进遗传算法对小波神经网络参数进行寻优,旨在解决小波神经网络参数敏感问题,进而提高诊断算法收敛速度和故障诊断准确率。仿真结果表明:与传统小波神经网络诊断方法相比,利用遗传算法改进的小波神经网络方法诊断正确率高达91%,提高了10个百分点。

The running state of the secondary circuit or operating mechanism of vacuum circuit breakers can be reflected by the characteristics of current curves.Firstly,three kinds of common faults,including core blockage,abnormal voltage(too high or too low)and breakdown,are simulated in laboratory,and a fault current curve characteristic library is established.Secondly,based on the property that the product of energy density in the inherent mode function of the fault current signals after ensemble mode decomposition and its corresponding average period is a constant,an improved empirical mode decomposition method(IEMD)is proposed to extract the current eigenvalues of the opening and closing coils,which are used as the input sample set of the neural network.On this basis,a circuit breaker fault diagnosis method is proposed by combining the improved genetic algorithm(GA)and wavelet neural network(WNN).This method uses the improved genetic algorithm to optimize the parameters of the neural network in order to solve the problem of parameter sensitivity of the wavelet neural network,thus improving the convergence speed of the diagnosis algorithm and the accuracy of fault diagnosis.Simulation results show that compared with the traditional neural network diagnosis method,the proposed fault diagnosis method has a diagnostic accuracy of 91%,increasing by 10 percentage point.