Abstract: To adapt to customized and quickly responded marketing demands,the facility layout problems which take the finite transport capacity of the material handling system into consideration are becoming increasingly critical.Aiming at the optimization of Pick-up/Drop-off points (P/D points) layout of intelligent workshops with unidirectional and multiple loops Automatic Guided Vehicle (AGV),the simulation-based optimization method combining intelligent algorithm and derivative-free search approach was put forward.By taking the average total transport cost as the objective function and performance index such as P/D points geometry locations and average congestion time of AGV as the constraints,an optimization model was established.A simulation-based Genetic Mesh Adaptive Direct Search algorithm (GMADS) that embedded Genetic Algorithm (GA) was used to solve the model in which the objective function had no closed mathematical expression.Based on the simulation information and the type of the problem,a method was proposed to improve the quality of solutions and the efficiency of the algorithm,which ameliorated the search directions of the algorithms.A group of contrast experiments and a practical case showed that the embedded GMADS algorithm for solving the P/D points layout problem was effective,efficient and practical.

Key words: pick-up/drop-off points layout, simulation-based optimization, genetic mesh adaptive direct search, derivative-free optimization

摘要： 为了适应定制化、快速响应的市场需求,考虑物料储运系统有限运载能力的设施布局问题日益重要。针对具有单向多重封闭回路自动导引小车(AGV)的智能车间制造单元上料与下料(P/D)口布局问题,研究了将智能搜索算法与无导数直接搜索算法相结合的仿真优化方法。建立以最小化平均运输总成本为目标函数,以P/D口几何位置和AGV平均拥堵时间等车间运行过程性能指标为约束的优化模型;针对该模型的目标函数没有封闭的数学表达形式的特点,设计了一种基于仿真的遗传网格自适应直接搜索算法(GMADS)求解该模型;提出一种基于仿真信息与问题特征的方法,改善算法搜索方向,提高解的质量及算法的效率;通过设计对比实验和实际智能车间的应用案例,验证了所提优化算法求解车间单元P/D口布局问题的有效性、优越性及其应用价值。

关键词: 上下料口布局, 仿真优化, 遗传网格自适应搜索算法, 无导数优化