高蛋白大豆肥效预测模型的正交多项式回归分析

Orthogonal Polynomial Regression Analysis of Fertilizer Efficiency Prediction Model for High Protein Soybean

为优化高蛋白大豆生产中的肥料配施,采用多元正交多项式回归预测模型来反映氮磷钾肥施用量和大豆产量的依存关系。结果表明:随氮、钾肥施用量增加,大豆产量先增后降,大豆产量与磷肥用量正相关,肥效大小表现为氮肥>磷肥>钾肥;肥效预测模型方差估计值σ=31.2616,预测高蛋白大豆产量U^的95%的置信区间为U±1.96σ=U±61.2727,效果较好;经模型拟合,氮、钾、磷施肥量分别为56.39,70.22和140.00kg·hm^-2时,可得大豆最高产量3365.73kg·hm^-2。多元正交多项式回归模型对预测多因素施肥水平下高蛋白大豆产量值有实际参考价值。

In order to optimize the fertilizer application in high-protein soybean production,multivariate orthogonal polynomial regression was used in this paper to model the dependence between fertilizer amount and soybean yield.The results showed that with the increase of nitrogen and potassium fertilizer application,soybean yield first increased and then decreased,and with the increase of phosphorus fertilizer,soybean yield showed a positive correlation line trend,and the effect size was nitrogen>phosphate>potassium.The variance estimate of regression model established by orthogonal polynomial was■=31.2616,and the 95%confidence interval of high-protein soybean yield per hectare was■±1.96σ=U±61.2727,with good effect.According to the model fitting,the maximum yield of soybean was 3365.73 kg·hm^-2,when N,K and P were 56.39,70.22 and 140.00 kg·hm^-2,respectively.The multivariate orthogonal polynomial regression model has practical reference value for predicting the yield of high-protein soybean under multi-factor fertilization.