基于振动时频图像和D-S证据理论的内燃机故障诊断

IC engines fault diagnosis based on vibration time-frequency image and D-S evidence theory

为抑制循环波动性对内燃机故障诊断结果的影响,引入D-S证据理论,提出一种基于内燃机振动时频图像、局部非负矩阵分解、BP神经网络和D-S证据理论的内燃机故障诊断新方法。首先采用平滑伪魏格纳分布(SPWVD)方法对8种不同气门状态的缸盖振动信号进行分析得到振动时频图像,然后用局部非负矩阵分解(LNMF)方法提取时频图像的特征参数并组成训练集和测试集,用得到的训练集对BP神经网络进行训练,再把测试集输入到训练好的BP神经网络,将输出的结果转化为基本概率赋值,用Deng加权平均证据合成规则对同种状态下不同图像的证据进行合成,并利用合成后的结果进行诊断分类。实例分析结果表明,基于振动时频图像和D-S证据理论的内燃机故障诊断方法可以有效抑制内燃机循环波动性对诊断结果的影响,能够准确诊断不同类型的气门故障。

D-S evidence theory is introduced into the IC engine fault diagnosis to suppress the influence of the IC engine cycle variation. A new method for IC engine fault diagnosis based on time-frequency images of vibration, local non-negative matrix factorization, BP neural network and D-S evidence theory was proposed. First, the vibrating time-frequency images were obtained by analyzing the vibration signals of the cylinder head with eight different valve states using the Smoothed Pseudo-Wigner Distribution （SPWVD） method. Then the local non-negative matrix factorization （LNMF） method was used to extract the characteristic parameters of the time-frequency images and form the training set and test set. The training set was used to train the BP neural network. The results of the trained neural network were transformed into the basic probability assignment. The evidence of three images in the same state was fused by Deng＇s weighted average evidence rule, and the fusion results were used for classification. The experimental results show that the proposed fault diagnosis method based on vibration time-frequency images and D-S evidence theory can effectively restrain the influence of cycle variation of IC engine on diagnosis result and can accurately diagnose different types of valve failures.