电静压伺服机构故障诊断中的主成分分析方法应用  

Application of Principal Component Analysis in Fault Diagnosis of Electro-Hydrostatic Actuators

作　　者：肖雪 赵守军[1] 陈克勤[1] 张朋[1] 刘璐[1] 

机构地区：[1]北京精密机电控制设备研究所,北京100076

出　　处：《导弹与航天运载技术》2019年第1期94-100,共7页Missiles and Space Vehicles

摘　　要：研究采用主成分分析方法(Principal Component Analysis,PCA)诊断电静压伺服机构的故障。为了诊断电静压伺服机构的故障,测试健康工况和故障工况下的阶跃特性,选取位置指令、反馈、速度和电机电流等参数作为多分析变量。建立健康工况的主元模型,得到平方预测误差(SquaredPredictionError,SPE)统计量及Hotelling-T2(T2)统计量及其阈值。在此基础上,对故障工况进行主成分分析,得到SPE和T2及其相对于健康工况的阈值对比情况,从而判断其是否故障。试验研究了增压油箱漏气及油滤堵塞两种故障工况,作动器速度均发生了明显变化。但相比传统的单一速度值判别,主成分分析得到的SPE可以将故障差异的数字量化程度进一步提高一个数量级,并且不同故障模式下的T2差异程度也不相同。研究结果表明,主成分分析方法可以有效地识别电静压伺服机构的故障及其特征,可以作为预测健康管理中的一种可选信号融合工具。Application of Principle Component Analysis (PCA) to the fault diagnosis of Electro-Hydrostatic Actuators (EHA) is studied. Application of Principle Component Analysis (PCA) to the fault diagnosis of Electro-Hydrostatic Actuators (EHA) is studied. Step responses are tested for both healthy and faulty conditions, with position commands, position feedbacks, velocity and the electric motor currents selected as multiple parameters for analysis. The principle component model is first built for a healthy condition to obtain both Squared Prediction Error (SPE) and Hotelling-T2 (T2 ) statistics as well as their thresholds. Thereafter, the PCA is implemented for a faulty condition to get its SPE, T2 and their comparison to the previous healthy condition to discriminate differences for diagnosis. Experiments are studied for 2 faults, including the gas leakage of the pressurized oil reservoir and the filter blockage in flow passage. It is shown that, under both 2 faults, their actuator velocity changes obviously. However, compared to the conventional judgment only by the velocity value, the SPE by PCA can further upgrade the faulty magnitude by one numerical order, at the same time, the T2 discrepancy modes of the faults are also different. It is demonstrated that PCA can effectively discriminate the faults and their characteristics for EHAs, and can be as an optional data fusion tool for their Prognostics and Health Management (PHM).

关 键 词：电静压伺服机构 故障诊断 主成分分析 

分 类 号：V448.15[航空宇航科学与技术—飞行器设计]