【作者】 薛尧予；

【导师】 王建林；

【作者基本信息】 北京化工大学 ， 控制理论与控制工程， 2010， 博士

【摘要】 发酵工程是生化工程和现代生物技术及其产业化的基础。在发酵工程领域,为了提高发酵水平和生产率,更多的研究工作集中在菌种的筛选和改造上。尽管现代生物技术的发展,在基因工程和代谢工程领域内有了长足的进展,通过诱发变异、基因重组和培养能够得到高产菌株,然而,通过优化模型和控制以使发酵过程产品生产最优仍是发酵工程领域中存在的主要问题之一,因此对生物发酵过程模型优化及优化控制的研究日益受到重视。粒子群优化算法原理简单、易于实现,且适合于复杂优化问题的求解,因此,将粒子群优化算法引入发酵领域进行模型参数估计为发酵过程模型优化提供了有效途径。发酵过程的优化控制目标多种多样(最大生产率、最大终止时刻产量或最高原料转化率等),发酵领域中处理多目标问题的传统方法(目标加权合并、目标转化为约束等)实施困难且易丢失非凸目标函数最优解以至决策失误。基于粒子群优化的多目标算法由于在搜索中具有多向性和全局性,同时可以处理所有类型的目标函数和约束,因此非常适合求解发酵过程中复杂的多目标优化控制问题。本文在分析现有粒子群算法研究现状的基础上,对标准粒子群算法在寻优过程中容易过早收敛、陷入局部最优的现象进行研究,应用能量守恒原理,通过引入粒子最差位置提出了一种群能量守恒粒子群优化算法。该算法根据粒子内能进行动态分群,对较优群体采用引入最差粒子的速度更新策略,加快较优群体收敛速度；对较差群体采用带有惩罚机制的速度更新策略,补偿较优群体速度降低产生的整群能量损失,避免算法陷入局部最优。典型优化问题的仿真结果表明,该算法具有更强的全局搜索能力和更快的收敛速度。对多目标进化算法在寻优过程中的收敛性和分布性问题进行研究,提出一种群能量守恒多目标粒子群优化算法。该算法在粒子速度和位移计算中引入粒子群体能量守恒机制,并将该机制同非支配排序方法、自适应网格机制以及精英保留策略进行有机结合,提高粒子搜寻能力,避免陷入次优非支配前沿。将该算法和非支配排序遗传算法分别作为子种群进化规则,构造基于种群间优劣互补的多目标协同进化算法。与经典多目标进化算法的比较测试结果表明,所提算法具有更好的解分布性和收敛性。在发酵过程优化控制方法研究上,针对批次发酵过程模型不准确和过程参数不稳定特点,利用批次流加过程中的反复迭代特性提出一种用于批次流加发酵过程的批次间协同优化控制方法。该方法将群能量守恒粒子群算法、多目标粒子群算法和批次间优化控制有机地结合起来,用上一轮批次流加发酵过程的数据进行过程模型参数辨识,并将更新的过程模型用于新一轮发酵过程中进行操作条件优化。基于工业酵母发酵过程仿真模型进行批次间协同优化控制方法实验,结果表明该方法有效地解决了批次流加发酵过程中的模型不准确和状态不稳定问题,实现了批次流加发酵过程优化控制。本文所提出的群能量守恒粒子群算法具有更强的全局搜索能力和更快的收敛速度；所提出的群能量守恒多目标粒子群优化算法具有很好的收敛性和分布性；基于群能量守恒粒子群算法和进化多目标协同算法的批次间协同优化控制方法为生物发酵过程优化控制提供了有效途径。更多还原

【Abstract】 Fermentation is the basis of bioengineering as well as modern biology technology and bioengineering industrialization. In the fermentation engineering, many studies have focused on the selection and transformation of the bacteria in order to improve the level and productivity of fermentation. Modern biotechnology has made significant progress in genetic engineering and metabolic engineering fields. Strains can be high yield by induced mutation, gene recombination, and cultured. But, getting the best products by optimizing the model and control of fermentation process is still one of the main problems existing in the fermentation engineering. Therefore, control optimization of microbial fermentation process studies received increasing attention. Particle swarm optimization method is simple, easy to implement, and suitable for complex optimization problems. Therefore, using Particle Swarm Optimization algorithm for optimization of fermentation process model has become an effective way to improve the fermentation process optimal control level. In the optimization of fermentation process control, it is needed to achieve multi-objective control tasks (such as the highest yield, the shortest, the minimum) to improve production efficiency and economic efficiency. In fermentation area, traditional methods of deal with multi-objective problems (such as weighted combining of objectives and transforming objectives into constraints) are difficult to implement and easy to lose the optimal solution of non-convex objective function, resulting in the decision-making mistakes. Multi-objective evolutionary algorithm based on PSO is very suitable for solving the complex multi-objective optimal control problem of the fermentation process, because it is a global optimization algorithm, and can handle almost all types of objective function and constraints.Based on the analysis of existing research on Particle Swarm Optimization algorithm and the research on the phenomenon of easy to premature convergence and sink into local optimum, which usually occurs in the optimization process of particle swarm optimization algorithm, a Swarm Energy Conservation Particle Swarm Optimization (SEC-PSO) is proposed. SEC-PSO, which is designed with the concept of energy conservation, can solve the problem of premature convergence frequently appeared in standard PSO algorithm by partitioning its population into several sub-swarms adaptively according to the energy of the swarm. The simulation results of typical optimization problems show that the algorithm has better global search capability and faster convergence.We studied the convergence and distribution of the optimization process of multi-objective evolutionary algorithms, and proposed a Swarm Energy Conservation Multi-objective Particle Swarm Optimization (SEC-MOPSO) algrothm. In this algrothm, a Swarm Energy Conservation mechanism is used. The mechanism is combined with non-dominated sorting method, adaptive grid mechanism, and elitist strategy to improve the search capabilities of particles and avoid falling into the second-best non-dominated front. We structured a co-evolution algorithm based on complementary strengths and weaknesses among populations. In this algorithm, evolution rules of sub-population are Non-dominated sorting Genetic algorithm (NSGA) and SEC-MOPSO respectively. The simulation results show that the proposed algorithms have better distribution and convergence performence of solution than classic multi-obective evolutionary algorithm.In the research of fermentation process optimal control method, a run-to-run optimization exploits the repetitive nature of fed-batch processes in order to deal with the optimal problems of fed-batch fermentation process with inaccurate process model and unsteady process state. The kinetic model parameters, which are used in the operation condition optimization of the next run, is adjusted by calculating time-series data got from real fed-batch process in the run-to-run optimization. The simulation is taken based on industrial yeast fermentation process simulation model. The results show that this strategy can adjust its kinetic model dynamically and overcome the instability of fed-batch process effectively.SEC-PSO proposed in this paper has a strong global search capability and fast convergence. Pareto optimal solutions of SEC-MOPSO and CEMO have good convergence and distribution. Run-to-run strategy with SEC-PSO and CEMO provides an effective method to control optimization of fed-batch fermentation process.更多还原