连续投影算法的润滑油中含水量的近红外光谱分析

Analysis on near infrared spectroscopy of water content in lubricating oil using successive projections algorithm

应用近红外光谱技术结合连续投影算法(SPA)实现了油中含水量的分析。对57个油样进行光谱扫描,通过比较不同预处理方法,以相关系数(R)和均方根误差(RMSE)作为模型评价指标,建立油中含水量预测的全波段偏最小二乘法(PLS)模型。同时应用SPA提取有效波长,作为PLS的输入变量,建立了SPA-PLS模型。结果表明经连续投影算法提取24个特征波长建立的模型,所用变量数仅占全波段的4.68%,SPA-PLS优于全波段的PLS模型,其对验证集样本进行预测的相关系数和均方根误差分别为0.994 4和5.455 1×10-5,获得了满意的预测精度。说明应用光谱技术检测油中含水量是可行的,并能获得满意的预测精度,为进一步应用光谱技术进行油中其他污染物的在线监测提供了新的方法。

Near infrared （NIR） spectroscopy combined with successive projections algorithm （SPA） was investigated for determination of water content in oil. A total of 57 oil samples were scanned, the correlation coefficient （R） and root mean square error （RMSE） were used as the model evaluation indices, the full-spectrum partial least squares （PLS） model was developed for the prediction of water content in oil after the performance comparison of different pretreatments. Simultaneously, successive projections algorithm was applied for the extraction of effective wavelengths, the selected effective wavelengths were used as the inputs of partial least squares （PLS）. The results indicated that a total of 24 variables, only 4.68 percent in the full spectrum, selected by SPA are employed to construct the model with 0.994 4 as the correlation coefficient and 5.455 1×10-5 as the root of mean square error of validation set, PA-PLS model is better than full-spectrum PLS model. An excellent prediction precision was achieved. In conclusion, successive projections algorithm is a powerful way for effective wavelength selection, and it is feasible to determine the water content in oil using near infrared spectroscopy and SPA-PLS, and an excellent prediction precision was obtained. This study supplied a new and alternative approach for further application of near infrared spectroscopy in on-line monitoring of contamination such as water content in oil.