论文部分内容阅读
为利用红外光谱分析技术快速、准确测定在用航空润滑油酸值,将小波变换用于润滑油红外光谱预处理中,结合均值中心化法,提取光谱有效信息建模。光谱的小波变换选择符合光谱特征的db4小波为基函数,在分解尺度9下进行光谱分解,利用软阈值法滤除各层干扰噪声,重构消噪信号。该光谱预处理法与传统的Savitzky-Golay平滑导数结合均值中心化法相比,滤噪效果好,有效压缩了建模数据量。采用偏最小二乘法,选择最佳主因子数5,建立酸值模型,并对10个在用航空润滑油油样进行了分析。
In order to quickly and accurately determine the acid value of aviation lubricating oil by using the infrared spectrum analysis technology, the wavelet transform is used in the oil infrared spectrum pretreatment, and the mean value centering method is combined to extract the spectral effective information. The wavelet transform of spectrum is selected as the basis function of db4 wavelet which is in accordance with the spectral features. Spectral decomposition is carried out at the decomposition scale of 9, and the interference noise of each layer is filtered by soft thresholding to reconstruct the de-noising signal. Compared with the traditional Savitzky-Golay smooth derivative and mean-centering method, the spectral pretreatment method has good filtering effect and effectively reduces the amount of modeling data. Partial Least Squares (LDLS) method was used to choose the best principal factor of 5, and the acid value model was established. The oil samples of 10 aviation lubricants were analyzed.