基于实数编码遗传算法的发酵过程优化控制

针对谷氨酸发酵过程的复杂性,利用人工神经元网络建立了谷氨酸发酵过程的动态数学模型,并利用实数制编码的遗传算法,采用过程整体优化的思路,以发酵过程中糖转化为谷氨酸的转化率为优化目标对发酵过程的多个操作变量同时进行优化,得到各个操作变量的最优控制轨迹.考虑到发酵过程中流加操作的重要性,将流加操作开始和结束的时间作为控制变量进行了优化.与产酸率为优化目标的仿真结果比较表明,该方法使发酵过程转化率有很大提高,且产酸率也接近于后者的最优产酸率.

Optimal Control Based on Real-Coding Genetic Algorithm for Fermentation Process

Considering the complexity of the fermentation process of glutamic acid,a dynamic neural network model is developed for the fermentation process.To optimize simultaneously the operating variables based on the optimal object of the conversion rate to glutamic acid from carbohydrate,the real-coding genetic algorithm (RCGA) is applied to the multivariable optimal control for the whole fermentation process,thus finding out the optimal control trajectories of different operating variables.Considering the importance of fed-batch operation in the fermentation process,the starting/ending time of fed-batch operation are taken as variables and then optimized.Comparing the simulation results of the optimization of output rate of glutamic acid with the optimization proposed for the control of the fermentation process,the latter can improve greatly the conversion rate mentioned above with the output rate approximate to that the simulation results revealed.