基于稳健估计的硫酸钙光谱值抗差能力分析

Analysis of Robustness of Spectral Values of Calcium Sulfate Based on Robust Estimation

硫酸钙在光谱观测过程中存在随机噪声,随机噪声会携带到观测的光谱数据中,对数据拟合造成影响,进而影响硫酸钙含量的测定。本研究应用最小二乘法和稳健估计法对光谱数据进行处理,选取硫酸钙5个吸收位置572.5 nm、1487.9 nm、1935 nm、2100.6 nm、2337.7 nm处的光谱反射率与对应的硫酸钙浓度进行相关性分析,得出相关性最大的光谱反射率;将这些反射率与硫酸钙浓度做回归分析,分析数据在存在粗差的情况下,最小二乘法回归分析和稳健估计法线性拟合的效果差异。结果表明:当光谱数据中存在粗差时,稳健估计法的抗差效果比最小二乘法更加明显,可用来建立较为可靠的模型。

There is random noise in the process of spectral observation of calcium sulfate,which will be carried into the observed spectral data and affect the data fitting,thus affecting the determination of calcium sulfate content.The least square method and robust estimation method were used in spectral data processing.The spectral reflectance at five absorb positions of calcium sulfate such as 572.5 nm and 1487.9 nm,1935 nm,2100.6 nm and 2337.7 nm were selected to analyze correlation with corresponding calcium sulfate concentration,and the maximum spectral reflectance of relevance were obtained.The reflectivity and the concentration of calcium sulfate were regressed,the effect of the linear fitting between the least squares regression analysis and robust estimation method were analyzed under the datagross error.The results show that the robust estimation method is more effective than the least square method,and can be used to establish a reliable model.