为提高梨可溶性固形物含量(soluble solids content, SSC)的近红外光谱模型的精度和稳定性, 以160个梨样品为实验对象, 分别对原始光谱、 多元散射校正(MSC)和标准正态变量变换(SNV)处理后的光谱, 经无信息变量消除算法(UVE)挑选后, 再结合遗传算法(GA)和连续投影算法(SPA), 筛选梨可溶性固形物的近红外光谱特征波长。将筛选后的波长作为输入变量建立梨可溶性固形物的最小二乘支持向量机(LS-SVM)模型。结果表明经过SNV-UVE-GA-SPA从全波段3112个波长中筛选出的30个特征波长建立的梨可溶性固形物LS-SVM模型效果最好, 该模型的预测集相关系数(Rp)和预测均方根误差(RMSEP)分别为0.956和0.271。该模型简单可靠, 预测效果好, 能满足梨的可溶性固形物含量的快速检测, 为在线检测和便携式设备开发提供了理论基础。

To improve the precision and robustness of the NIR model of the soluble solid content (SSC) on pear. The total number of 160 pears was for the calibration (n=120) and prediction (n=40).Different spectral pretreatment methods, including standard normal variate (SNV) and multiplicative scatter correction (MSC) were used before further analysis. A combination of genetic algorithm (GA) and successive projections algorithm (SPA) was proposed to select most effective wavelengths after uninformative variable elimination (UVE) from original spectra, SNV pretreated spectra and MSC pretreated spectra respectively. The selected variables were used as the inputs of least squares-support vector machine (LS-SVM) model to build models for determining the SSC of pear. The results indicated that LS-SVM model built using SNVE-UVE-GA-SPA on 30 characteristic wavelengths selected from full-spectrum which had 3112 wavelengths achieved the optimal performance. The correlation coefficient (Rp) and root mean square error of prediction (RMSEP) for prediction sets were 0.956, 0.271 for SSC. The model is reliable and the predicted result is effective. The method can meet the requirement of quick measuring SSC of pear and might be important for the development of portable instruments and online monitoring.