摘要
提出了一种基于可见—近红外光谱分析技术快速鉴别发动机润滑油含水量的新方法。采用美国ASD公司的便携式光谱仪对5种不同含水量的润滑油进行了光谱分析,各取30个样本数据。采用偏最小二乘回归法对样本数据进行处理,随机抽取100个样本用于建模,余下的50个样本用于模型验证。试验结果表明,模型的相关系数达到了0.874206,预测均方根误差RMSEP为0.013565,说明提出的方法具有较好的鉴别作用,为发动机润滑油含水量的快速鉴定提供了一种新方法。
A new fast measurement method of water content of engine lubricant based on visible and near infrared （Vis/NIR） spectroscopy was put forward.150 samples from 5 different water contents level lubricant samples were collected for Vis/NIR spectral scanning.Partial least squares （PLS） analysis was applied as calibration method after preprocessing stage to build the forecast model.The calibration set was composed of 100 samples,whereas 50 samples in the validation set.The results indicated that PLS model achieved the best prediction performance.A new process proposed in this paper for determination of water content of engine lubricant.
出处
《农机化研究》
北大核心
2010年第7期222-223,228,共3页
Journal of Agricultural Mechanization Research
基金
国家"十一五"科技支撑计划项目(2006BAD10A0403)
浙江省教育厅科技项目(20071275)
