特征波长筛选在近红外光谱测定梨硬度中的应用

为了提高应用近红外光谱分析技术快速测定梨硬度的精度和稳定性，该研究采用联合区间偏最小二乘和遗传算法（siPLS-GA）在校正模型中用来筛选特征光谱区域和波长，通过交互验证法确定模型的主成分因子数和筛选的波长，并以预测均方根误差（RMSEP）和相关系数（Rp）作为模型的评价标准。基于siPLS-GA的最优模型包含4个光谱区、96个变量和10个主成分因子。该模型结果显示：最佳预测模型相关系数（Rp）和RMSEP分别为0.9083和0.5573。研究结果表明，近红外光谱技术结合siPLS-GA建模用于无损、快速测定梨的硬度是可行的。

Application of characteristic wavelengths selection in determination of pear firmness by near infrared (NIR) spectroscopy

In order to improve the detecting precision and robustness in determination of pear firmness by the FT-NIR spectroscopy, in this research, Synergy interval partial least square coupled with genetic algorithm (siPLS-GA) was used to select the efficient spectral regions and wavelengths in calibrating model. The number of components and the number of variables were implemented by the cross-validation. The performance of the final model was evaluated according to the root mean square error of prediction (RMSEP) and correlation coefficient (R) in prediction and calibration sets. The optimal model based on siPLS-GA was obtained with 10 PLS factors, while 4 spectral regions and 96 variables were selected, respectively. The results of final model show that the optimal model can obtain correlation coefficient of 0.9083, and RMSEP of 0.1573 respectively by a prediction set. The research demonstrated that pear firmness could be determined by NIR spectroscopy technique is feasible, and siPLS-GA the superiority in calibrating model.