应用近似最优设计建立多元肥效模型的方法探讨

optimun design. Acta Agriculturae Universitatis Zhejiangensis,

本文简要介绍了D-最优设计的特点及其基本原理,分析了D-最优设计的试验处理点在因子区域上的布局与农化研究要求的不一致性,在适当放弃D-优良性的基础上提出了更适合农化研究要求的近似最优设计方案。报道了由饱和设计试验资料建立的肥效模型所存在的拟合效果良好而预报性质不良的问题,并提出了“扩点建模”方法使两者得到统一。

The basic principle of D-optimun design and its advantages were introduced briefly. But there were some problems when it was adopted in the experiment which was conducted for establishing the multiple fertilizers response function. For example, the disiribution of its treatment points in the factorial space was not harmonious with the demand of the experiment, so an approaching optimun design was recommended, which might be more consistent with the experimental demand although its D-optimun property was lost partially. It was reported that the multiple fertilizers response functions were of excellent goodness of fit, but showed poor estimation. A new method called 'augmenting points method' for establishing multiple fertilizers response function was introduced, which had given an effective result.