Liu Fei, Zhang Fan, Fang Hui, Jin Zong-Lai, Zhou Wei-Jun, He Yong
College of Biosystems Engineering and Food Science, Zhejiang University, Hangzhou 310029, China.
Guang Pu Xue Yu Guang Pu Fen Xi. 2009 Nov;29(11):3079-83.
Near infrared (NIR) spectroscopy combined with successive projections algorithm (SPA) was investigated for the fast and nondestructive determination of total amino acids (TAA) in oilseed rape leaves. Total amino acids are important indices of the growing status of oilseed rape. A total of 150 leave samples were scanned, the calibration set was composed of 80 samples, the validation set was composed of 40 samples and the prediction set was composed of 30 samples. The optimal partial least squares (PLS) model was developed for the prediction of total amino acids in oilseed rape leaves after the performance comparison of different pretreatments, including smoothing method, standard normal variate (SNV), the first derivative and second derivative. Simultaneously, successive projections algorithm was applied for the extraction of effective wavelengths (EWs), which were thought to have least collinearity and redundancies in the spectral data. The selected effective wavelengths were used as the inputs of multiple linear regression (MLR), partial least squares (PLS) and least square-support vector machine (LS-SVM). Then the SPA-MLR, SPA-PLS and SPA-LS-SVM models were developed for performance comparison. The determination coefficient (R2) and root mean square error (RMSE) were used as the model evaluation indices. The results indicated that both SPA-MLR and SPA-PLS models were better than full-spectrum PLS model, and the best performance was achieved by SPA-LS-SVM model with R2 = 0.983 0 and RMSEP = 0.396 4. An excellent prediction precision was achieved. In conclusion, successive projections algorithm is a powerful way for effective wavelength selection, and it is feasible to determine the total amino acids in oil-seed rape leaves using near infrared spectroscopy and SPA-LS-SVM, and an excellent prediction precision was obtained. This study supplied a new and alternative approach to the further application of near infrared spectroscopy in the response of stress and on-field monitoring of the growing oilseed rape.
研究了近红外（NIR）光谱结合连续投影算法（SPA）用于快速无损测定油菜叶片中的总氨基酸（TAA）。总氨基酸是油菜生长状况的重要指标。共扫描了150个叶片样本，校准集由80个样本组成，验证集由40个样本组成，预测集由30个样本组成。在对不同预处理方法（包括平滑方法、标准正态变量变换（SNV）、一阶导数和二阶导数）进行性能比较后，建立了用于预测油菜叶片中总氨基酸的最优偏最小二乘法（PLS）模型。同时，应用连续投影算法提取有效波长（EWs），这些波长在光谱数据中被认为具有最小的共线性和冗余性。将所选的有效波长用作多元线性回归（MLR）、偏最小二乘法（PLS）和最小二乘支持向量机（LS-SVM）的输入。然后建立了SPA-MLR、SPA-PLS和SPA-LS-SVM模型进行性能比较。测定系数（R2）和均方根误差（RMSE）用作模型评估指标。结果表明，SPA-MLR和SPA-PLS模型均优于全光谱PLS模型，SPA-LS-SVM模型性能最佳，R2 = 0.983 0，RMSEP = 0.396 4，实现了优异的预测精度。总之，连续投影算法是一种有效的波长选择方法，利用近红外光谱和SPA-LS-SVM测定油菜叶片中的总氨基酸是可行的，并且获得了优异的预测精度。本研究为近红外光谱在油菜胁迫响应及田间生长监测中的进一步应用提供了一种新的替代方法。
