摘要
多目标优化是过程系统工程的重要课题,通常以加权或约束方式将其转换为单一目标,未能反映多目标间的复杂关系,不利于随时根据需求作出有效的决策。基于群智能的粒子群算法具有全局优化性能,且易于实现。为使其适于多目标优化,应拓展功能,实施改造。以Pareto支配概念评价种群个体的优劣,设计了确定局部最优点和全局最优点的操作。又利用各粒子的局部最优点信息进行速度更新,以加强种群的多样性,避免因早熟而陷于局部最优。还设置了外部优解库,并通过分散度计算,以适当的策略进行更新,使之逐步均匀地逼近于Pareto最优解集。由此构建一种多目标粒子群优化算法(multi-objective particle swarm optimization,MOP-SO),并用于补料分批生化反应器的动态多目标优化,取得了满意的结果。可基于所搜得的Pareto最优解集,分析目标间的关系,为合理决策提供有效的支持。经与NSGA-II比较,MOPSO算法具有更为优良的性能。
Multi-objective optimization is an important topic of process systems engineering. Through simply converting to a single goal, it often fails to reflect the more complex relationship between goals, and it is not conductive to effective decision-making at any time on demand. Swarm intelligence based particle swarm optimization （PSO） algorithm has good global optimization performance and can be easily implemented. To make it suitable for multi-objective optimization, PSO should be further rebuilt. Firstly, the concept of Pareto dominance was used to evaluate the fitness of particles, and two kinds of operation for determining the local and global optimal point respectively were designed. Secondly, velocity update strategy, utilizing all particles＇ local best information, was used to enhance the ability of global convergence. Thirdly, an external archive technique was set up, and through calculating the degree of dispersion, an appropriate update strategy was adopted to uniformly approximate the Pareto optimal solution set step-by-step. Finally, multi-objective particle swarm optimization （MOPSO） was proposed, and it was applied to dynamic multi-objective optimization of fed-batch bioreactor, the satisfactory solution was obtained. According to the obtained pareto optimal solution set, the relationship between goals could be analyzed further, which could contribute to rational and effective decision-making. Compared with NSGA-Ⅱ , MOPSO showed better performance.
出处
《化工学报》
EI
CAS
CSCD
北大核心
2007年第5期1262-1270,共9页
CIESC Journal
基金
国家自然科学基金项目(20276063)~~
关键词
多目标
粒子群算法
均匀逼近
PARETO最优集
补料分批生化反应器
动态优化
multi-objective
particle swarm optimization
uniform approximation
Pareto optimal solution set
fed-batch bioreactor
dynamic optimization
