基于粗糙集神经网络和振动信号的高压断路器机械故障诊断

Mechanical Fault Diagnosis of High Voltage Circuit Breakers Based on Rough Set Neural Networks and Vibration Signals

为了准确检测出高压断路器的机械故障类型,该文提出一种基于本征模态边际谱能量与粗糙集神经网络相结合的高压断路器振动信号故障诊断方法。首先将断路器的振动信号经过经验模态分解(EMD),得到若干个本征模态函数(IMF),对各个IMF分量进行希尔伯特(Hilbert)变换得到Hilbert边际谱,求取Hilbert边际谱的二次方得到Hilbert边际谱能量作为特征向量。基于粗糙集理论对特征向量进行属性约简分析,从而建立简单明了的决策表,根据决策表规则建立径向基函数(RBF)神经网络故障模型。实验结果表明,该方法能有效对高压断路器的机械故障类型进行分类。

In order to accurately detect the type of mechanical failure of high-voltage circuit breakers,A fault diagnosis method for high-voltage circuit breaker vibration signals based on the combination of the intrinsic mode function(IMF)spectrum energy and the rough set neural network is proposes.Firstly,the circuit breaker vibration signal is decomposed by empirical mode decomposition(EMD),and then several IMF is obtained.Hilbert transform is performed on each IMF component to obtain the Hilbert marginal spectrum.Its square is called the marginal spectrum energy as the feature vector.Based on the rough set theory,the attribute reduction analysis is performed on the eigenvectors to establish a simple and clear decision table.The radial basis function(RBF)neural network fault model is established according to the decision table rules.Experimental results show that this method can effectively classify the types of mechanical faults of high-voltage circuit breakers.