求解复杂耦合问题的多系统优化方法

由于复杂耦合问题具有多系统、多目标、多约束、多尺度和不确定等特点, 急需一种求解此类问题的高效 智能优化方法. 为此, 借鉴多种群进化算法的智能平行特征, 利用种群间进化信息的继承和交互作用, 提出一种多 系统优化方法. 首先以子种群来代表子系统的优化环境, 通过子系统内的进化操作求解各自的优化子问题; 然后通 过子系统间的迁移操作, 即利用变量共享、目标函数和约束条件的相似程度来实现子系统间的信息迁移与反馈, 加 速整个问题的全局优化; 最后将该方法应用到基准函数和具有多系统优化特征的三级供应链网络, 仿真实验表明 所提出的方法可行且有效.

Multi-system optimization method for complex coupling problems

Because of the characteristics of multi-system, multi-objective, multi-constraint, multi-scale and uncertainty for complex coupled problems, an efficient intelligent optimization method is urgently needed to solve such problems. Based on the parallel feature of multi-population evolutionary algorithms, a multi-system optimization method is proposed taking advantage of inheritance and interaction of evolution information among different subpopulations. First, the subpopulation is used to represent the optimization environment of a subsystem, and within-subsystem evolution operations are used to solve their respective optimization subproblems. Second, cross-subsystem migration operations that combine variable sharing, and the similarity of objective functions and constraints, are used to realize information transfer and feedback between the subsystems, which accelerates the global optimization for the whole problem. Finally, the proposed multisystem optimization method is applied to the benchmark functions and three-echelon supply chain networks composing of multiple systems and multiple objectives. The simulation results show that the proposed method is feasible and effective.