基于发动机磨损变化的时间序列模型,提出基于模型的特征根位置诊断方法、Green函数诊断方法和残差方差变化的3种发动机磨损状态诊断方法,并通过仿真和发动机磨损实测数据对上述诊断方法进行了验证。特征根方法和Green函数方法是从产生观测数据的系统稳定性度研究磨损状态变化,而残差方差方法则是从观测数据的统计特性来研究系统的变化。特征方程根位置诊断方法对于判定磨损状态在理论上给出了严格的阈值,Green函数方法同样可以用来诊断系统的稳定性,但这一方法对于轻微的磨损现象表现出不敏感,而参差方差方法对系统的变化最为敏感。
New fault diagnosis methods, the characteristic root position diagnosis method, Green function diagnosis method and changing of residual sum of squares were promoted based on the time series model. The effect of those methods was validated by the simulate model data and real experimental data of diesel engine. Characteristic root position and Green function diagnosis method are in view of the satiability of the system, however, the residual sum of squares method is in view of the statistics of the data which come from the wear system. Characteristic root position method can give the strict threshold in theory. Green function diagnosis method also can be used to judge the system satiability, but this method is not sensitive to the changing of the light wear, the residual sum of squares method is the most sensitive of the system of the three methods.