基于局部线性模型树模型的柴油机进气系统漏气和堵塞故障诊断

Fault Diagnosis of Intake System of Diesel Engine Based on LOLIMOT

柴油机进气系统漏气和堵塞等性能退化型故障将会导致排放恶化和经济性下降。针对进气系统具有较强的非线性,难以建立精确的数学模型问题,提出基于局部线性模型树(LOLIMOT)模型的故障诊断方法,为在线故障诊断提供一种思路。提取进气压力波的一次谐波信号幅值与相位、充量系数为故障特征,采用LOLIMOT方法对充量系数、进气压力波动幅值与相位信号建立基于发动机转速和进气密度的参考模型;采用奇偶方程生成3个残差信号,计算各个残差信号的阈值;分析残差信号和故障类型的映射关系。结果表明:所建进气压力波动幅值LOLIMOT模型、进气压力波动相位LOLIMOT模型、充量系数模型验证数据与模型仿真数据具有较高线性相关度;采用充量系数、进气压力波幅值与相位为故障特征值建立的LOLIMOT模型生成残差信号的故障诊断方法,可诊断进气系统漏气故障和中冷器堵塞故障。

Leakage and blocking of intake system of diesel engine may lead to the deterioration of the emissions and fuel economy. An accurate model of intake system is difficult to be established since it is a strong nonlinear system. Thus, a fault diagnosis method based on local linear model tree （LOLIMOT） is proposed for on-line fault diagnosis. The amplitude and phase of intake pressure fluctuation signal and vol- umetric efficiency are selected as the fault characteristic parameters. LOLIMOT is used to establish the reference models of the parameters. The models are regarded as ~r function of engine speed and intake density. Three residual signals are generated by using parity equation for calculating their threshold val- ues. The mapping relationship between the residual signal and the fault type is analyzed. Experimental results show that the proposed models give good prediction of fault characteristic parameters, the proposed method can be to diagnose the leakage of engine intake system and the blocking of intercooler.