基于时域能量划分和PSO-SVM的发动机故障诊断

Engine Fault Diagnosis Based on Time-domain Energy Division and PSO-SVM Algorithm

针对发动机缸盖振动信号激励源的时序性,提出了一种基于时域能量划分和粒子群优化-支持向量机算法(PSO-SVM)的发动机故障诊断方法。将振动信号按曲轴转角进行时域能量划分并将各段时域信号能量组成的向量作为故障分类的特征向量,应用不同优化技术的支持向量机(SVM)算法对发动机配气机构故障进行识别。诊断结果表明:时域能量划分结合PSO-SVM在小样本的情况下,能够对既定机型的配气机构和点火系常见故障进行识别。

In view of the time sequence of excitation source of engine cylinder head vibration signals, an engine fault diagnosis method based on time-domain energy division technique and particle swarm optimization-sup- port vector machine （PSO-SVM） algorithm is proposed. The vibration signals are divided into time-domain energy according to crankshaft angle, the vectors composed of each time-domain signal energy are taken as the eigenvectors of fault classification, and the faults of engine valve train are identified by SVM algorithm with different optimization techniques. The diagnosis results show that the combination of time-domain energy division with PSO-SVM algorithm can identify the common faults of valve train and ignition system small sample set. of the specific type of engine under the condition of