基于多元非线性回归分析的马铃薯加工品质特性预测

为快速准确检测马铃薯加工品质,对希森3号及希森6号马铃薯的加工品质指标干物质和还原糖含量在马铃薯内部的分布规律进行研究。通过试验得出,两种马铃薯干物质含量在马铃薯中呈与马铃薯形状相似的椭球分布,在中心部位有最小值。采用拟牛顿算法和通用全局优化算法,结合对试验数据进行多元非线性回归分析,得出希森3号和希森6号马铃薯干物质含量关于检测点坐标值x、y和z的回归模型,决定系数分别为0.909 9和0.912 3,均能有效预测马铃薯干物质含量。马铃薯还原糖含量在马铃薯中心位置含量最高,由中心向表皮还原糖含量逐渐降低,在马铃薯茎部的还原糖含量低于马铃薯顶部关于中心对称位置的还原糖含量。两种马铃薯还原糖含量的多元非线性回归模型,决定系数分别为0.833 6和0.824 6,可预测马铃薯内部各点还原糖含量。通过对试验数据归纳分析得出,马铃薯靠近表皮位置干物质含量高,还原糖含量低,适合加工薯片薯条。

Potato Processing Quality Characteristics Prediction Based on Multivariate Nonlinear Regression Analysis

Aimming to accurately detect potato processing quality, the distribution of processing quality index dry matter and reducing sugar in potato were investigated. It was concluded that potato dry matter had the ellipsoid distribution in potato and the minimum value was in the center part. The experimental data was analyzed by using quasi-Newton algorithm combined with universal global optimization method, and the regression model of dry matter content of Xisen 3 and Xisen 6 potato about testing point coordinates x, y, and z was received. The determination coefficients of regression models about dry matter content of Xisen 3 and Xisen 6 potato were 0.9099 and 0.9123, respectively, and regression models can effectively predict the potato dry matter content. The content of reducing sugar in the center of potato was the highest, and the reducing sugar content from the center to the epidermis was decreasing. The reducing sugar content in the stem of potato was lower than that on the top of potato. The determination coefficients of multivariate nonlinear regression models about two kinds of potato reducing sugar content were 0.8336 and 0.8246, respectively. The regression models could predict the contents of reducing sugar in potato. According to the analysis of experimental data, the potato close to the epidermis was higher in dry matter and lower in reducing sugar, which was suitable for processing potato chips and French fries. The average content of dry matter of Xisen 6 potato is more than 20%, and the average content of reducing sugar is less than 0.1%, which is suitable for processing. This research can help processors for effective utilization of potato for various types of processing products viz., chips, French fries and flour, and provide a theoretical basis for NIR nondestructive detection of potato processing quality.