提高二次回归肥料试验成功率及函数典型率的研究

Study on Increasing the Successful and Typical Probability of Quadratic Function Regression in Fertilizer Experiments

在配方施肥工作中,无论采用何种试验设计进行的二次回归肥料试验,均会产生非典型效应函数,无法用于最佳施肥量的确定。为提高二次回归肥料试验成功率及函数典型率,选用D-最优设计进行水稻和玉米试验,设置2次重复,并采用小区多向优化组合排列法、多种数学模型回归统计、误差数据校正等措施优化试验结果,使96次试验的成功率达60%。

The fertilizer formula of crop response function to fertilizer will produce most proportion non-typical fertilizer effect functions in different experiment design and quadratic function regression, it can not determine optimum rate of application. To solve the problems, in D-saturated optimized design, using replication, multi-direction plot arrangement, different mathematical pattern regression analysis, correction of error data for maximization of fertilizer experiment, the success ratio of 96 fertilizer experiments in rice and corn amounts to 60%.