EEMD和SVM在发动机故障诊断中的应用

针对发动机缸盖振动信号的非线性非平稳特征,提出一种总体平均经验模态分解(EEMD)和支持向量机相结合的信号分析及故障诊断方法,该方法利用EEMD算法以及IMF序列和原始振动信号之间的相关系数,有效放大故障诊断特征向量的差异。对原始振动信号进行EEMD分解,得到各阶特征模态函数(IMF),求各阶IMF分量对应于原始信号的相关系数并组成故障分类特征向量。分别将IMF相关系数法和IMF能量分布法得到的特征向量作为输入,建立BP神经网络和支持向量机,判断发动机工作状态和故障类型。分析表明,对IMF求相关系数的方法简便易行,能有效放大不同工况下特征向量的差异,结合支持向量机能够对既定机型的配气机构和点火系常见故障进行准确识别。

Application of EEMD and SVM in Engine Fault Diagnosis

For the non‐linear and non‐stationary characteristics of cylinder head vibration signal ,the method of ensemble em‐pirical mode decomposition (EEMD) combined with support vector machine (SVM ) was proposed .The difference of feature vectors was amplified effectively by using EEMD algorithm and correlation coefficient of intrinsic mode function (IMF) se‐quence with original vibration signal .The IMFs were acquired by decomposing the original vibration signal with EEMD and then the feature vectors of fault classification was formed by calculating the correlation coefficients of IMF with original signal . Taking the feature vectors calculated with IMF correlation coefficient and energy distribution method as the input ,BP neural network and SVM models were established to analyze engine working status and fault type .The result shows that the method of IMF correlation coefficient is feasible and can magnify the difference between feature vectors .With the help of SVM ,the common faults of valve mechanism and ignition system can be identified accurately .