基于偏最小二乘回归分析发动机异常磨损的判断  

Engine Abnormal Abrasion Estimation Based on Partial Least-Squares Regression

作　　者：蔡晓光[1] 刘玉兵[2] 王晓东 

机构地区：[1]中国矿业大学机电学院,江苏徐州221000 [2]徐州空军学院军交运输指挥系,江苏徐州221000 [3]济南空军装备部,山东济南250000

出　　处：《润滑与密封》2010年第2期69-70,75,共3页Lubrication Engineering

摘　　要：针对发动机状态监测中油液光谱分析样本数据少、元素种类多及各元素之间存在相关性的特点,提出采用偏最小二乘回归方法处理油液光谱分析数据。分析了偏最小二乘回归方法的特点,阐述了其基本原理和算法步骤;将偏最小二乘回归方法应用于发动机油液光谱数据的处理和分析,求出了发动机磨损状态判断的回归模型;应用此回归模型判断了EQ6BT柴油发动机的异常磨损。结果表明,应用此回归模型可以直观地判断发动机是否存在异常磨损。Engine oil spectrum analysis has some defect of insufficiency in sample data,excessive of elements and correlation between different elements.Partial least-squares regression （PLSR） was presented to process engine oil spectrum analysis data.The characteristic of PLSR was discussed.The principle and algorithm were explained.The PLSR method was applied to the disposal and analysis of engine spectrum analysis data,the regression model that can be used to estimate the abrasive state of engine was got.The abnormal abrasion of EQ6BT diesel engine was estimated with the model.The result shows that the abnormal abrasion of engine can be found by the PLSR model.

关 键 词：偏最小二乘回归 油液光谱分析 异常磨损 发动机 

分 类 号：TH117.1[机械工程—机械设计及理论]