基于振动信号的发动机综合故障诊断

Comprehensive Fault Diagnosis of Engine Based on Vibration Signal

为提高发动机综合故障诊断效果,对发动机不同部位、不同工况下的振动信号采用多种方法进行特征提取。将提取出的多类型特征参数结合转速因子组成综合特征向量,作为训练样本和测试样本,输入到多隐含层BP神经网络中进行故障诊断。试验结果表明:基于振动信号的多特征值和多隐含层BP神经网络的方法能够实现对发动机的综合故障诊断。

To improve the effect of comprehensive diagnosis of engine faults,the paper utilizes multiple methods to extract features of vibration signals at different parts of an engine under diverse operating conditions,combines those features with speed factors to form feature vectors,and input them as training and test samples into BP neural network with multiple hidden layers.Experimental results show that this new method,which integrates multiple eigenvalues of vibration signals with multiple hidden layers of BP neural network,can realize a comprehensive diagnosis of engine faults.