基于小波能量与神经网络的断路器振动信号识别方法

高压断路器出现机械故障不仅会引起振动冲击事件的时间漂移,还会引起时域波形中一些波峰幅值的变化。依据同一类型断路器振动信号相似的特点,在对高压断路器故障振动信号进行特征分析的基础上,提出了一种识别高压断路器振动信号的新方法:将小波包提取算法和径向基神经网络模式识别功能相结合,利用小波包分解与重构原理将断路器合闸振动信号分解到不同频段中,提取每个频带能量作为断路器状态监测的特征向量,作为径向基神经网络的输入向量;基于径向基神经网络的故障诊断方法在系统参数未知的情况下自动建立动态模型,对于线性系统和非线性系统都有很好的跟踪能力,通过实验室断路器典型合闸振动信号的监测及识别分析验证了该方法的有效性。

Circuit breaker vibration signal recognition based on wavelet energy and neural network

Mechanical failure of high voltage circuit breaker may cause the time shift of vibration impulse events and bring the change of some waveform amplitude peaks in time domain.As the vibration signals of circuit breakers in same type are similar,a recognition method of high voltage circuit breaker vibration signals based on its characteristic analysis is presented,which combines the wavelet packet extraction algorithms and neural network-based radial pattern recognition function together.The breaker-close vibration signals are decomposed into different frequency bands by wavelet packet decomposition and reconstruction and the energy of each band is used as the eigenvector of circuit breaker condition monitoring,as well as the input vector of RBF(Radial Basis Function) neural networks.RBF neural network based fault diagnosis system builds the dynamic model automatically,which has good tracking ability for both linear and nonlinear systems with unknown parameters.Its validity is proved by the monitoring and recognition of typical breaker-close vibration signals in laboratory.