基于遗传算法与迭代动态规划混合策略的青霉素发酵过程优化控制

Optimization control strategies combined genetic algorithms and iterative dynamic programming for penicillin fermentation processes

对于青霉素发酵过程优化控制问题,考虑到遗传算法难以处理模型的不确定性以及迭代动态规划搜寻效率过度依赖初始给定轨迹问题,论文提出了一种两者结合的混合策略,使用遗传算法对青霉素补料发酵过程最优补料速率进行初始求解,结果作为迭代动态规划算法给定参考值,进行二次优化。遗传算法优化得出的青霉素发酵产量是8.02 g/L,经过迭代动态规划算法的二次优化之后,青霉素的产量为8.27 g/L,优化效果均比常规等级补料优化控制策略给出的结果有所提高。给定初始参考控制轨迹之后,迭代动态规划算法的运行时间由243 s降为179 s,算法运行效率大大提高。可见混合优化控制策略既可以提高迭代动态规划寻优效率,同时也在一定程度上克服了遗传算法对于过程模型不确定性敏感的问题。混合优化控制策略的思想可以为类似微生物发酵过程的优化控制提供参考。

Regarding optimization control of penicillin fermentation processes, difficulty in dealing with model uncertainty and dependency on initial trajectory in searching of iterative dynamic programming approaches are concerned. In response, this paper introduces a combined optimization control strategy, which uses genetic algorithms to achieve initial optimum feeding rate of penicillin fermentation processes before employing iterative dynamic programming to re-optimize the processes. Penicillin fermentation production under genetic algorithm optimization is 8.02 g/L. The production of penicillin was 8.27 g/L after re-optimization by iterative dynamic programming, the optimization effect of both these two methods are better than the normal feeding strategy. Given the initial reference control trajectory, running time of iterative dynamic programming decreased from 243 s to 179 s, which indicates that efficiency of the IDP algorithm is greatly increased. It is shown that this method can not only improve the efficiency of iterative dynamic programming optimization but also overcome the limitations that genetic algorithms are sensitive to process model uncertainties. Combined optimization control strategy can provide a reference for similar optimization control in microbial fermentation processes.