润滑油红外光谱去噪预处理方法

采用小波包变换软阈值去噪方法对润滑油的红外光谱进行去噪预处理,结合偏最小二乘法进行定量预测,并与常用的平滑去噪方法进行对比。结果表明,小波包变换能有效地去除红外光谱的噪声,以此为基础建立的润滑油酸值模型的预测精度高于常用的平均平滑法和 Savitzky-Golay卷积平滑法这 2种平滑去噪方法的预测精度。

Pretreatment of Noise Removal in Infrared Spectrum of Lubricating Oil

The pretreatment of noise removal in the infrared spectrum was made by using the wavelet packet transform, andthen the quantitative prediction was made combining with the partial least square and comparing to the two commonlymethods. Results reflected that wavelet packet transform could remove the noise in infrared spectrum effectively. The pre.diction accuracy by the acid model we established based on that was more efficient than the average smoothing method andSavitzky-Golay method.