能谱熵向量法及粒子群优化的RBF神经网络在高压断路器机械故障诊断中的应用

Application of Energy Spectrum Entropy Vector Method and RBF Neural Networks Optimized by the Particle Swarm in High-voltage Circuit Breaker Mechanical Fault Diagnosis

高压真空断路器是电力系统开关设备中极其重要的一种高压电器,而高压断路器故障中80%是由于机械特性不良造成,为此通过小波包变换对高压断路器机械振动信号进行了分析,以信号的能谱熵作为特征输入向量,建立了粒子群优化(PSO)径向基函数(RBF)神经网络的高压断路器故障识别系统模型,最后对实际高压断路器振动信号进行获取分析并得到结果。实验结果表明,高压断路器正常信号能谱熵向量各元素分布比较均匀;而故障信号所得能谱熵向量各元素变化较大且有一定变化规律;粒子群优化后的RBF网络模型在正确率、精度等方面高于传统神经网络模型。实验结果表明该方法用于高压断路器的故障诊断是可行的,并且可以为断路器的故障诊断提供更好的理论依据。

High voltage vacuum circuit breaker is an important high voltage appliance of power system switchgears, and 80 of failure of the high voltage circuit breaker is due to its poor mechanical properties. Thus, using the signal energy spectrum entropy as character input vectors, we established a fault identification system model of the high voltage circuit breaker, which was based on particle swarm optimization by radial basis function neural network, and analyzed vibration signals of high-voltage circuit breakers through the wavelet packet. Moreover, we obtained and analyzed the actual vibration signals of high-voltage circuit breaker. Experimental results show that each element of the spectrum entropy vector of normal signal of high-voltage circuit breakers is evenly distributed, while the elements of fault signal spectrum entropy vector remarkably regularly change. The accuracy and precision of RBF network model which is optimized by particle swarm are higher than those of the traditional neural network model. Experimental results also show that the method for high voltage circuit breaker fault diagnosis is feasible and effective, and can provide a better theoretical basis for fault diagnosis of circuit breakers.