Abstract: To address the distributed precast scheduling problem,considering the characteristics of mixed interruptible and non-interruptible operations,mixed serial and parallel operations,as well as constraints on order acceptance for factories in precast component production,and the varying impact of geographical locations in the transportation process,a sequence-based mixed-integer nonlinear programming model was established to minimize production-transportation costs and inventory-delay penalties.Given the complexity of the problem,a hybrid intelligent optimization algorithm based on adaptive large neighborhood search was proposed.In this algorithm,an ordinal-based vector encoding and decoding method was first designed,followed by a combination of dynamic neighborhood extraction heuristic and taboo search algorithms to enhance the quality of initial solutions.Diverse neighborhood structures and a multi-strategy fusion approach were introduced to further enhance the solution quality and solution efficiency to prevent premature convergence to local optima.Finally,the effectiveness of the proposed algorithm was validated through extensive experimentation.

Key words: distributed precast production scheduling, just-in-time, mixed integer nonlinear programming, adaptive large neighborhood search

摘要： 针对分布式预制构件生产调度优化问题,考虑预制构件生产中可中断与不可中断工序混合、串并行工序混合、工厂接受工件数量约束等特点,以及生产与运输环节中不同地理位置对调度结果的影响,以最小化生产与运输成本和库存与拖期惩罚费用为目标,建立了一类基于排序的混合整数非线性规划模型。鉴于问题的复杂性,提出一种基于自适应大邻域搜索的混合智能优化算法。在该算法中,首先设计了一种基于序数的向量组编码解码方式,随后,结合动态邻域提取启发式算法和禁忌搜索算法以提高初始解质量。为防止算法过早收敛而陷入局部最优,引入多种邻域结构并运用多策略融合思想进一步提升问题解的质量和求解效率。最后,通过计算实验验证了所提算法的有效性。

关键词: 分布式预制构件生产调度, 准时制, 混合整数非线性规划, 自适应大邻域搜索