作者：刘鑫¹, 贾云献¹, 崔心瀚², 王强¹

作者单位：1. 陆军工程大学石家庄校区, 河北 石家庄 050003;
2. 白城常规兵器试验中心, 吉林 白城 137001

关键词：油液信息;特征提取;BP神经网络;状态评估

摘要：

基于油液信息的设备健康状态评估在工程实践中具有重要的应用，为选择最优的油液信息特征提取方法并实现柴油机的健康状态评估，分别研究描述油液信息和设备健康状态关系的油液浓度分析模型、梯度分析模型和比例分析模型，并以BP神经网络为评估方法，实现基于油液信息的柴油机状态评估，通过评估结果对3种油液信息模型的特点进行总结，并对其各自特点的原因进行分析。结果表明，油液梯度模型能充分利用油液中的状态信息，评估效果好。同时，对柴油机的状态评估研究也具有一定的工程实用价值。

Research of feature extraction for lubricant data and its application of diesel engine condition evaluation

LIU Xin¹, JIA Yunxian¹, CUI Xinhan², WANG Qiang¹

1. Army Engineering University of PLA, Shijiazhuang 050003, China;
2. Baicheng Ordnance Test Center, Baicheng 137001, China

Abstract: Equipment healthy state evaluation based on lubricant analysis play an important role in engineering, for selecting the optimal information processing method of lubricant data and realizing the state evaluation of engine, respectively to describe the relationship between the lubricant data and equipment status with the concentration model, gradient model and proportion analysis model in lubricant, the state evaluation of engine was realized based on the lubricant analysis data with the BP neural network model. The characteristics of three kinds of lubricant information processing were summarized through the evaluation results and their reasons of characteristics were analyzed respectively. The results show that the analysis model based on the metal gradient in lubricant can make full use of the state information of the lubricant, which has the best evaluation effect.At the same time, the engine state evaluation also have some practical value in engineering.

Keywords: lubricant data;feature extraction;BP neural network;condition evaluation

2018, 44(6): 121-128  收稿日期: 2017-09-20;收到修改稿日期: 2017-11-09

基金项目: 国家自然科学基金项目（71401173）

作者简介: 刘鑫(1989-),男,山东淄博市人,博士研究生,研究方向为可靠性、装备故障诊断及寿命预测。

参考文献

[1] WANG W, ZHANG W. A model to predict the residual life of aircraft engines based on oil analysis data[J]. Nav Res Log, 2005, 52:276-284.
[2] WANG W, HUSSIN B. Plant residual time modeling based on observed variables in oil samples[J]. J Oper Res Soc, 2009, 60:789-796.
[3] DAVID V, LIBOR Z, ONDǐEJ P. Contribution to system failure occurrence prediction and to system remaining useful life estimation based on oil field data[J]. Proc IMechE Part O:J Risk and Reliability, 2015, 229(1):36-45.
[4] WANG W B, HUSSIN B, JEFFERIS T. A case study of condition based maintenance modelling based upon the oil analysis data of marine diesel engines using stochastic filt ering[J]. Int. J. Production Economics, 2012(136):84-92.
[5] 孙磊, 贾云献, 蔡丽影. 基于油液光谱分析和粒子滤波的发动机剩余寿命预测研究[J]. 光谱学与光谱分析, 2013, 33(9):2478-2482.
[6] 刘学工. 基于油液监测技术的重型车辆故障诊断研究[D].长春:长春理工大学, 2013.
[7] DAVID S, CANH L. Methodology and framework for predicting helicopter rolling element bearing failure[J].IEEE Transactions on Reliability, 2012(61):846-857.
[8] 任国全, 郑海起, 张英堂,等. 基于油液分析的自行火炮发动机磨损状态监测研究[J]. 兵工学报, 2002, 23(1):6-9.
[9] 陈丽. 基于状态的维修模型及应用研究[D]. 石家庄:军械工程学院, 2008.
[10] FARZANEH A, JAN L. Remaining useful life prediction of grinding mill liners using an artificial neural network[J]. Minerals Engineering, 2013(53):1-8.
[11] 付强. 基于铁谱分析的船用齿轮箱磨损状态评估方法研究[D]. 哈尔滨:哈尔滨工业大学, 2013.