遗传算法联合LS-SVM的苹果原醋成分定量分析

利用近红外光谱技术对苹果原醋中的重要指标进行定量分析，并进行模型优化以提高性能。采用遗传偏最小二乘法（GA-PLS）提取的特征波长作为最小二乘支持向量机（LS-SVM）的输入变量，先后建立苹果原醋中总酸、可溶性固形物的近红外定量模型，并与建立的偏最小二乘（PLS）模型结果进行比较。用决定系数（R2）、预测均方根误差（RMSEP）以及相对分析误差（RPD）对模型进行评价，确定最佳建模方法。结果表明，相比于PLS模型，总酸及可溶性固形物指标的LS-SVM定量模型的R2、RMSEP以及RPD值均有更好的表现，且在进行独立测试集验证时，LS-SVM模型的预测精度也明显优于PLS模型。说明遗传算法联合LS-SVM建立的定量模型有很高的准确度及稳定性，可以应用于苹果原醋总酸和可溶性固形物含量的快速检测。

Quantitative analysis of apple vinegar compositions based on genetic algorithm combined with LS-SVM

The compositions of apple vinegar were analyzed quantitatively by near infrared spectroscopy technology, and the model was optimized to improve the performance. The characteristic wavelengths extracted by genetic algorithm partial least squares (GA-PLS) as least squares support vector machines (LS-SVM) of the input variables, NIR quantitative models of total acid and soluble solid in apple original vinegar were established, and the models were compared with partial least squares (PLS) model results. The established models were evaluated using R2, RMSEP and RPD to determine the optimum modeling method. The results showed that R2, RMSEP, RPD and prediction accuracy in independent test set of LS-SVM quantitative model of total acid and soluble solid had better performance than PLS models. The quantitative model established by genetic algorithm combined with LS-SVM had high accuracy and stability. It could be used in the rapid detection of total acid and soluble solids content in the apple vinegar.