论文部分内容阅读
为了解决定量关系不符合 Beer定律的红外光谱波长选择问题 ,提出了基于校正模型定量误差估计的波长选择方法。该方法从光谱点噪声对校正模型的误差影响角度出发 ,对每个光谱点的定量性能进行评估 ;利用这个评估结果对利用波长子集建立的模型误差进行了理论估计。这种方法在对光谱点的误差估计过程中考虑了光谱数据的非线性特点 ,因而能够处理非线性光谱数据的波长选择问题。仿真实验表明 :该方法的波长选择是成功的 ,并且其效果优于基于线性运算的遗传算法
In order to solve the problem that the quantitative relationship does not accord with Beer’s law, the wavelength selection method based on the quantitative error estimation of the calibration model is proposed. The method evaluates the quantitative performance of each spectral point from the point of view of the influence of spectral point noise on the error of the calibration model. By using this evaluation result, the model error established by using wavelength subset is theoretically estimated. This method takes into account the nonlinear characteristics of the spectral data in the estimation of the error of spectral points and thus can deal with the wavelength selection problem of nonlinear spectral data. The simulation results show that the wavelength selection of this method is successful and its effect is better than the linear arithmetic based genetic algorithm