This paper focuses on modeling and optimizing design for continuous conversion of glycerol to 1,3-propanediol by Klebsiella pneumoniae, wherein dilution rate of the feed medium is constantly changed with fermentation time. A nonlinear three-phase hybrid system is first established to describe this bioconversion process. With mean productivity of 1,3-propanediol as objective function and dilution rate as design variable, we then formulated the optimization design problem as a discrete-valued optimal control problem, which is eventually approximated to a large-scale parameter optimization problem via time transformation and multiple shooting approach. A multi-strategy competition artificial bee colony algorithm is developed to solve this large-scale problem. Computer simulations show the rationality of the proposed hybrid system and the feasibility of dynamic dilution rate. Numerical results indicate that, on the one hand, mean productivity in continuous culture is improved obviously using dynamic dilution rate; on the other hand, the developed numerical algorithm is efficient and valid for this complex applied optimization problem.

本文重点研究肺炎克雷伯菌将甘油连续转化为 1,3-丙二醇的建模和优化设计，其中补料培养基的稀释率随发酵时间不断变化。首先建立非线性三相混合系统来描述这种生物转化过程。以1,3-丙二醇的平均生产率为目标函数，稀释率为设计变量，将优化设计问题表述为离散值最优控制问题，最终通过时间变换近似为大规模参数优化问题和多种射击方式。为了解决这个大规模问题，开发了一种多策略竞争人工蜂群算法。计算机模拟显示了所提出的混合系统的合理性和动态稀释率的可行性。数值结果表明，一方面，采用动态稀释率，连续培养的平均生产率显着提高；另一方面，开发的数值算法对于这个复杂的应用优化问题是有效的。