特征谱区筛选在近红外光谱检测茶叶游离氨基酸含量中的应用研究

在近红外光谱快速检测茶叶游离氨基酸含量过程中，为了提高检测的精度和稳定性，研究利用特征谱区结合偏最小二乘法建立预测模型。研究分别尝试联合区间偏最小二乘法和遗传偏最小二乘法等特征谱区筛选方法，通过交互验证法确定偏最小二乘模型的主成分因子数和筛选区间，以预测均方根误差RMSEP和相关系数R作为模型的评价指标。试验结果表明：两种方法建立模型的预测能力都好于传统PLS模型；利用联合区间偏最小二乘法建立的预测模型最佳，预测时的相关系数(R)和预测均方根误差(RMSEP)分别为0.9542和0.2560。研究结果表明，近红外光谱结合特征谱区筛选方法可以快速准确地测定茶叶中游离氨基酸含量。

Selection of the efficient wavelength regions in FT-NIR spectroscopy for determination of free amino acid content in tea

In determination of free amino acid content in tea by FT-NIR spectroscopy, in order to improve its precision and robustness, it attempted to build predictive model by the efficient spectral region selection and PLS algorithm. In this research, synergy interval partial least-squares (siPLS) and genetic algorithm partial least-squares (GA-PLS) were applied to select the efficient spectral regions. The number of PLS components and the number of intervals were optimized according to cross-validation in calibration set. The performance of the final model was evaluated according to root mean square error of prediction (RMSEP) and correlation coefficient (R) in prediction and calibration sets. Experimental results showed that the two methods were able to produce better prediction models in relation to the full-spectrum model, and the performance of siPLS model was better than GA-PLS. The optimal model was achieved with R= 0.9542 and RMSEP =0.2560 in prediction set. This study demonstrated that NIR spectroscopy with efficient selection wavelength regions algorithm could be used successfully to analyze free amino acid content in tea, and revealed superiority of siPLS algorithm in contrast with other multivariate calibration methods.