面向多目标柔性作业车间调度的强化学习NSGA-II算法

针对非支配排序遗传算法 (NSGA-II, non-dominated sorting genetic algorithm II)在求解柔性作业车间多目标优化调度问题中多样性不足、易于早熟与局部收敛的缺点,提出一种基于强化学习的改进NSGA-II算法(RLNSGA-II, reinforcement learning non-dominated sorting genetic algorithm II)。为避免NSGA-II陷入局部收敛问题引入双种群进化策略,利用性别判定法将种群拆分为两个种群,并在进化过程中采用不同的交叉变异算子,增加算法的局部和全局搜索能力;为解决NSGA-II精英策略造成多样性不足的问题,融合多个多样性度量指标,利用强化学习动态优化种群迭代过程中的拆分比例参数以保持多样性,改善算法收敛性能。最后通过Kacem标准算例进行了仿真实验与性能分析,验证了RLNSGA-II的有效性与优越性。

Reinforcement learning NSGA-II for multi-objective flexible job shop scheduling

Non-dominated sorting genetic algorithm II (NSGA-II) has the shortcomings of insufficient diversity, prematurity and local convergence in solving the multi-objective optimal scheduling problem in flexible job shop. In this study, an improved NSGA-II algorithm based on reinforcement learning (RLNSGA-II) is proposed. To avoid NSGA-II to fall into the problem of local convergence, a two-population evolution strategy is introduced. The sex determination method is used to split the population into two populations, and different cross mut-ation operators are used in the evolution process to increase the local and global search capabilities of the algorithm. In order to solve the problem of insufficient diversity caused by the NSGA-II elite strategy, multiple diversity metrics are integrated, and reinforcement learning is used to dynamically optimize the split ratio parameters in the population iteration process to maintain diversity and improve algorithm convergence performance. Finally, simulation experiments and performance analysis are carried out through Kacem standard calculation examples, verifying the effectiveness and superiority of RLNSGA-II.