遗传算法在正交试验设计最优条件选择中的应用

APPLICATION ON GENETIC ALGORITHM IN SELECTING THE BEST EXPERIMENTAL CONDITION FOR ORTHOGONAL EXPERIMENTAL DESIGN

目的 :探讨正交试验设计中最优条件选择的方法。方法 :本文利用胃蛋白酶生产过程中质量控制的正交试验结果 ,以二次响应面回归模型为目标函数 ,用遗传算法搜索最优试验条件。结果 :二次响应面回归模型有统计学意义 (F=16 0 .0 ,P<0 .0 0 0 1,R2 =98.6 1 ) ,遗传算法确定的最优试验条件为 :水解温度 4 6 .7℃、水解 3.75 h、加 3% HCl、烘房温度 6 4 .1℃ ,该条件的精度较直接法高 ,此时胃蛋白酶中残留蛋白的预测量为 0 .0 0 0 1mol。结论 :模型与遗传算法结合 ,为正交试验设计最优条件的确定提供了新的方法。

Objective:To explore a method to determine the best experimental condition of the response variable for orthogonal experimental design.Methods:For the data of orthogonal experiment about measuring residual protein in the production of pepsin,on the basis of target function--the quadratic response surface regression model,the best experimental condition was searched by genetic algorithm (GA).Results:The results showed that the quadratic response surface regression model was significant( F=160.0,P<0.0001,R 2=98.61%) .The best experimental condition by GA was that with 46.7℃ of the hydrolysis temperature,3.75 hours of the hydrolysis time,3.0% HCl and 64.1℃ of the bake-house's temperature.The results showed GA was more accurate than the traditional method,while GA could predict the value of the residual protein,which reduced to 0.0001 mol.Conclusion:The response surface regression model and GA combined to provide a new method for determining the best experimental condition of the response variable for orthogonal experimental design.