近红外离散波长光谱基线漂移校正方法研究

发布时间：2018-03-18 23:31

本文选题：离散波长光谱　切入点：基线漂移校正　出处：《光谱学与光谱分析》2014年10期 　论文类型：期刊论文

【摘要】：微分是(近)红外光谱多元分析校正中最常使用也是最有效的光谱基线漂移校正方法。由于数据数目较少及相邻数据在光谱意义或数学意义上缺乏连续性,微分不能直接用于离散波长光谱消除基线漂移。为此,提出了一种结合插值拟合和微分校正离散光谱基线漂移的新方法。思路是采用三次样条插值法对离散波长光谱进行拟合,然后对拟合光谱进行Savitaky-Golay卷积求导,再从微分光谱中取出对应于原离散波长光谱数值的数值,构成离散波长光谱的微分光谱,从而实现离散波长光谱的基线漂移校正。通过分别由模拟离散波长光谱数据和实际的离散波长光谱数据建立多元校正模型检验新方法效果。采用ABC干粉灭火剂和土壤的近红外光谱数据及性质建立了PLS和MLR模型。结果表明,新方法能有效消除离散波长光谱的基线漂移对多元分析校正产生的不利影响,明显地提高了多元分析校正模型的准确性,对改善离散波长光谱仪器分析准确度具有重要的理论意义和实际应用价值。
[Abstract]:Differentiation is the most common and effective method for correction of spectral baseline drift in multivariate analysis correction of near infrared spectroscopy. Because of the small number of data and the lack of continuity of adjacent data in spectral or mathematical sense, Differential can not be directly used to eliminate baseline drift in discrete wavelength spectrum. A new method combining interpolation fitting and differential correction for baseline drift of discrete spectrum is proposed. The idea is to use cubic spline interpolation method to fit discrete wavelength spectrum. Then the derivative of the fitting spectrum is derived by Savitaky-Golay convolution, and then the value corresponding to the original discrete wavelength spectrum is extracted from the differential spectrum, which constitutes the differential spectrum of the discrete wavelength spectrum. In order to realize the baseline drift correction of discrete wavelength spectrum, the multivariate calibration model was established by simulating discrete wavelength spectrum data and actual discrete wavelength spectral data respectively. The new method was verified by using ABC dry powder fire extinguishing agent and the real discrete wavelength spectrum data. PLS and MLR models were established for the near infrared spectral data and properties of soil. The new method can effectively eliminate the adverse effect of the baseline drift of discrete wavelength spectrum on the correction of multivariate analysis, and improve the accuracy of the multivariate analytical correction model. It has important theoretical significance and practical application value to improve the analytical accuracy of discrete wavelength spectrometer.
【作者单位】：北京化工大学材料科学与工程学院;
【基金】：国家科技支撑计划课题(2011BAE11B00)资助
【分类号】：TH744.1

【参考文献】