Abstract: The monitoring methods were used separately to detect the faults in car dumper hydraulic systems based on physical model and statistical model. The monitoring variables were divided into some blocks according to the working mechanism in each stages of car dumper hydraulic systems, which makes the causal relationship of variables in each blocks more explicit. Then a MPCA model was built based on variables in each blocks. The multi-block MPCA model and all variables MPCA model were applied to leakage fault monitoring. The results show that the multi-block MPCA model is more sensitive to small leakage and has a higher fault detection rate.

Key words: physical model, multiway principal component analysis(MPCA), multi-block, fault detection, car dumper hydraulic system

摘要： 采用基于物理模型和统计模型的方法对翻车机液压系统故障进行监测。根据翻车机液压系统各阶段的工作机理将监测变量分块，使每一块中的变量间因果关系更加明确，再对各块分别建立多向主成分分析(MPCA)监测模型。将分块MPCA模型和全变量MPCA模型应用于压车缸泄漏故障的监测，结果表明分块MPCA模型对微小泄漏更加敏感，具有较高的故障识别率。

关键词: 物理模型, 多向主成分分析, 分块, 故障监测, 翻车机液压系统