多特性参数相结合的柴油机故障振动信号特征提取与诊断

Vibration Signal Feature Extraction and Diagnosis of Diesel Engine Faults with the Combination of Multi-Characteristic Parameters

本文基于缸盖振动信号的非平稳的特点,将经验模式分解和AR模型参数相结合,并兼顾了信号频谱的特点,提出了综合选取AR模型参数、信号频带能量和信号的质心频率与质心幅值等作为柴油机故障振动信号特征参量的方法;采用该方法对实测的S195柴油机的5种工况下缸盖振动信号样本提取了故障特征向量,基于支持向量机对柴油机故障进行诊断,故障诊断的正确率达到83%以上,验证了该方法的可行性。

Based on the non-stationary characteristics of the cylinder head vibration signals, combining empirical mode decomposition with autoregressive （AR） model parameters, and considering also the feature of signal spectrum, a method is proposed in this paper that comprehensively selects the AR model parameters, signal frequency band energy and the frequency and amplitude of signal gravity center, etc. as the characteristic parameters of fault vibration signals in diesel engines. With the method, the fault eigenvectors are extracted from the cylinder head vibration signal samples measured in S195 diesel engine at five working conditions. diagnosed based on support vector machine technique with an accuracy of over 83% method proposed. The faults of diesel engine are , verifying the feasibility of the method proposed.