基于反射光谱的润滑油磨粒含量检测实验研究

doi:10.3964/j.issn.1000-0593(2025)03-0826-10

摘要
参考文献
相关文章 (15)

全文: PDF (5893 KB)
输出: BibTeX | EndNote (RIS)

摘要：当发动机内部传动部件间发生磨损时，细微金属磨粒会脱落在设备内部构件之间，严重影响发动机的正常运行，甚至会引发严重事故，因此在线监测润滑油内磨粒信息是十分必要的。采用反射光谱定量分析，开展了润滑油磨粒含量检测实验，通过搭建反射光谱润滑油磨粒含量检测实验台，选用300目（50 μm）疲劳磨损颗粒和80目（175 μm）严重磨损颗粒共2种粒径大小的Fe粒和Cu粒，分别在可见光波段（450～760 nm）和紫外波段（200～435 nm），获取了润滑油磨粒浓度在6～15 μg·mL^-1范围内，以0.3 μg·mL^-1为间隔的31组反射光谱数据。针对反射光谱数据建立了偏最小二乘（PLS）线性模型，而该模型预测效果较差。又采用数据预处理校正方法对原始数据进行筛选校正，削减建模数据中的干扰因素，建立了PLS优化模型，却发现PLS优化模型虽然提高了预测效果，部分工况下的预测效果仍然较差，为进一步优化模型预测效果，建立了遗传规划模型和遗传规划-偏最小二乘（遗传规划-PLS）模型。最终得出以下结论：在PLS线性模型中，模型决定系数R²处于0.71～0.80范围内；在PLS优化模型中，模型决定系数R²处于0.80～0.94范围内；在遗传规划模型中，模型决定系数R²处于0.72～0.96范围内；在遗传规划-PLS模型中，模型决定系数R²处于0.84～0.98范围内。结果表明遗传规划-PLS模型预测效果最好。通过对润滑油磨粒含量反射光谱的研究，有望为发动机油液监测提供一种新的方法。

关键词：润滑油磨粒含量；反射光谱；光谱分析；模型优化

Abstract：When wear occurs between the internal transmission components of the engine, fine metal wear particles will fall off between the internal components of the equipment, which will seriously affect the normal operation of the engine and even cause serious accidents. Therefore, it is necessary to monitor the information of wear particles in lubricating oil online. In this paper, the detection experiment of lubricating oil wear particle content was carried out based on the quantitative analysis of the reflection spectrum. By building the experimental platform for detecting the wear particle content of lubricating oil by reflection spectrum, two kinds of Fe particles and Cu particles with particle sizes of 300 mesh (50 μm) fatigue wear particles and 80 mesh (175 μm) severe wear particles were selected. In the visible light band (450～760 nm) and the ultraviolet band (200～435 nm), 31 sets of reflection spectrum data of lubricating oil wear particle concentration in the range of 6～15 μg·mL^-1 with an interval of 0.3 μg·mL^-1 were obtained. Firstly, a partial least squares (PLS) linear model was established for the reflectance spectral data, but the prediction effect of the model was poor. Therefore, the data preprocessing correction method is used to screen and correct the original data. The interference factors in the modeling data are reduced, and the PLS optimization model is established. However, it is found that although the PLS optimization model improves the prediction effect, it is still poor under some working conditions. To further optimize the prediction effect of the model, a genetic programming model and a genetic programming-partial least squares (Genetic Programming-PLS) model are established. Finally, the following conclusions are drawn: the model determination coefficient R² is in the range of 0.71～0.80 in the PLS linear model, 0.80～0.94 in the PLS optimization model, 0.72～0.96 in the genetic programming model, and 0.84～0.98 in the genetic program-PLS model. The results showed that the genetic programming-PLS model had the best prediction effect. The study of the reflectance spectroscopy of wear particle content in lubricating oil is expected to provide a new method for engine oil monitoring.

Key words：Lubricating oil wear particles content; Reflectance spectroscopy; Spectroscopic analysis; Model optimization

收稿日期: 2024-03-12 修订日期: 2024-07-11

中图分类号:

TK314

基金资助: 国家重点研发计划项目（2021YFE0112800），陕西省企业院所联合重点专项（2023-LL-QY-29）资助

通讯作者: 周延 E-mail: yan.zhou@mail.xjtu.edu.cn

作者简介: 刘雪婧，女，2000年生，西安交通大学能源与动力工程学院硕士研究生 e-mail: liuxuejing@stu.xjtu.edu.cn

引用本文:

刘雪婧，崔洪帅，殷雄，马世一，周延，种道彤，熊兵，李锟. 基于反射光谱的润滑油磨粒含量检测实验研究[J]. 光谱学与光谱分析, 2025, 45(03): 826-835.
LIU Xue-jing, CUI Hong-shuai, YIN Xiong, MA Shi-yi, ZHOU Yan, CHONG Dao-tong, XIONG Bing, LI Kun. A Study on the Detection of Wear Particle Content of Lubricating Oil Based on Reflectance Spectrum. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(03): 826-835.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)03-0826-10 或 https://www.gpxygpfx.com/CN/Y2025/V45/I03/826