红外光谱结合化学计量学方法快速测定在用航空润滑油的酸值

Rapid determination of acid number in used aviation lubricating oil by FTIR spectrometry combined chemometrics

关键词:  在用航空润滑油  傅里叶红外光谱  酸值  偏最小二乘法  神经网络法 

Key words:  used aviation lubricating oil  acid number  Fourier transform infrared (FTIR)  partial least squares (PLS)  back propagation-artificial neural network(BP-ANN)

摘要:

研究采用傅里叶红外光谱快速测定在用航空润滑油酸值的方法。对光谱预处理方法进行研究,确定了S-G平滑和一阶导数的窗口宽度分别为19点和17点。在酸值相关基团分析的基础上,通过对比在用油和新油,确定了建模的光谱区间分别是3 597～3 202 cm-1、2 600～2 250 cm-1、920～800 cm-1。比较了偏最小二乘(PLS)法和神经网络(BP-ANN)法两种方法建立校正模型的效果,PLS模型的校正标准偏差(SEC)为0.004,预测标准偏差(SEP)为0.005。对PLS分析模型进行准确度分析,满足t检验的要求。研究结果表明,红外光谱法可准确测定在用航空润滑油的酸值,具有快速、无污染的特点。 

Abstract:

The method of acid number determination in used aviation lubricating oil by fourier transform infrared (FTIR) spectrometry was studied. Infrared spectra was pretreated by S-G smoothing and the first derivative, and the respective wave number of 19 point and 17 point were selected respectively. After analyzing the spectral absorption of acid radical and comparing the spectra of new oil and used ones, wave number ranges of 3 597~3 202 cm-1 , 2 600~2 250 cm-1 and 920~800 cm-1 were selected to establish model. Calibration model was built with chemometrics methods of partial least square(PLS) and back propagation-artificial neural network(BP-ANN) respectively. Standard deviation of calibration(SEC) of PLS model was 0.004, and standard deviation of prediction (SEP) was 0.005. The accuracy of the model was validated by means of t-test(α=0.05),and the result of t-test showed that acid number of used aviation lubricating oil could be determined quickly by FTIR spectrometry with advantages of high accuracy and no-pollution.