自适应混合EDA求解一类三阶段装配流水线调度问题

Adaptive hybrid estimation of distribution algorithm for solving a certain kind of three-stage assembly flowshop scheduling problem

针对生产过程中广泛存在的一类三阶段装配流水线调度问题,即带序相关设置时间的三阶段装配流水线调度问题,提出一种自适应混合分布估计算法,用于最小化平均完成时间和最大延迟时间的加权和。提出初始种群和初始概率分布模型生成机制,使概率分布模型能适当地积累较多优质解的信息,以提高AHEDA在进化初期的搜索能力。设计了基于信息熵的概率分布模型自适应更新机制和保留优良模式的新种群采样生成方法,增强了算法的全局搜索能力。引入基于Insert的邻域搜索来增强算法的局部搜索能力。最后通过仿真实验和算法比较验证了AHEDA的有效性。

Aiming at a certain kind of three-stage assembly flowshop scheduling problem which was Three-Stage As- sembly Flowshop Scheduling Problem with Scquence-Dependent Setup Times （TSAFSP_SDST）, an Adaptive Hy- brid Estimation of Distribution Algorithm （AHEDA） was presented to minimize the weighted sum of average com- pletion time and maximum tardiness. The generation mechanism of initial population and initial probability distribu- tion model were proposed to make probability distribution model accumulate the high quality solutions＇ information properly, which could improve the search ability of AHEDA at the initial stage of evolution. To enhance AHEDA＇s global search ability, the adaptive update scheme based on information entropy was designed for probability distribu- tion model, and the new population generation method was also constructed to keep excellent and good pattern. An Insert-based neighbor search was introduced to improve the local search ability. The effectiveness of the presented AHEDA was verified by computational experiments and comparisons.