基于支持向量机的汽车发动机故障诊断研究

Fault diagnosis for a car engine based on support vector machine

研究在汽车发动机不解体的情况下获取发动机缸盖表面的振动信号和上止点信号,按曲轴转角的周期对振动信号的时域能量进行合理分段;提取各段信号的时域能量作为发动机各缸状态的特征值。建立发动机各缸不同故障状态的训练集,用支持向量机的方法实现发动机故障模式的诊断与识别。研究结果表明,该方法对汽车发动机故障类型、故障位置的诊断识别具有重要工程意义。

The study was to obtain vibration signals and BDC signals on cylinder head surfaces without disassembling a car engine, then the energy of vibration signals in time domain was divided in to segments reasonably according to the crankshaft rotating period, and the energy in time domain of each segment was extracted as the state characteristic values of the engine cylinders. The training sets of the engine cylinders with different fault states were established. The support vector machine method was adopted to realize engine fault diagnosis and recognition. The results showed that the proposed method is helpful for diagnosis of engine fault type and fault location.