均匀设计及偏最小二乘回归法在高渗乳油研制中的应用

Using uniform design and partial least-squares regression to prepare hypertonic emulsifiable concentrates

以实例说明如何在高渗乳油制剂研制中应用均匀设计和偏最小二乘回归法。以制剂不含渗透剂与含有渗透剂时的润湿时间差值(Y1)、渗透时间差值(Y2)作为试验指标,以20%三唑磷高渗乳油配方研制为例,应用均匀设计法设计试验方案,对试验结果采用偏最小二乘回归法,以因变量最大值为优化方向,同时对两个试验指标进行回归建模,得到Y1与各渗透剂含量回归模型及Y2与各渗透剂含量回归模型,模型拟合的决定系数R2分别为0.9883、0.9978。最优目标函数值Y1、Y2分别为176.21s、137.18s,实测值分别为176.10s、137.11s。

This paper provides a case study illustrating the use of uniform design （UD） and partial least-squares （PLS） regression in developing an emulsifiable concentrate （EC） with high penetration ability. Differences in wetting time （Y1） and penetration time （Y2） between pesticide formulations with and without penetrating agent were used as an experimental indices, and different penetrating agent preparations of triazaphos 20% EC were used as factors in the study. UD protocols were used to design the experiments, and experimental results were processed using PLS regression. Maximum response values were used as the optimized regression direction. R^2 values of the models were 0.9883 and 0.9978, and the wetting and penetration times predicted by the models were 176.21 and 137.18 s. respectively, while actual test values were 176.10 and 137.11 s, respectively.