铁水运输调度双层多目标约束优化模型

为实现铁水运输作业排程与资源分配的协同优化，基于约束程序累积调度和字典序多目标优化理论，研究了铁水运输调度双层多目标约束优化方法.首先，基于铁水罐周转率最高和作业效率最高2个字典序优化目标，考虑作业时序、作业实施逻辑、铁水温降时限、铁水罐作业次数限制、资源容量限制和铁水罐资源池等约束条件，建立了上层的铁水运输作业排程约束优化模型；其次，以资源利用均衡度最高为目标，将作业实施唯一性和资源容量限制作为约束条件，建立了下层的铁水运输资源分配约束优化模型；最后，通过约束传播与多点构建性搜索的混合算法迭代求解整个模型.通过实例验证表明：设计的混合算法求得的铁水罐周转率目标和运输作业效率目标，比基本深度优先回溯算法分别提高了14.29%和60.53%；字典序多目标模型比加权和单目标模型求解效率和求解质量分别提高了20.3%和11.11%.

Double-Layer and Multi-objective Constraint Optimization Model for Transportation Scheduling of Molten Iron

In order to realize the collaborative optimization of operation scheduling and resource allocation in molten iron transportation, based on the theory of the cumulative scheduling with constraint programming and lexicographic multi-objective optimization, a double-layer and multi-objective constraint optimization method is explored for the transportation scheduling of molten iron. Firstly, setting the highest turnover rate of molten iron tanks and the highest operation efficiency as two lexicographic objectives, the upper-level constraint optimization model is built for molten iron transportation operation. In the model, the constraints are involved, such as operation sequence, operation implementation logic, time limit of molten iron cooling, limited operation times of molten iron tank, resource capacity limit, and resource pool of the molten iron tanks. Secondly, with the highest resource utilization balance, the lower-level constrained optimization model is established for resource allocation in molten iron transportation, in which the uniqueness of operation implementation and resource capacity are taken as constraints. Finally, the hybrid algorithm of constraint propagation and multi-point constructive search is developed to solve the whole model iteratively. The case study shows that, the turnover rate target and transportation efficiency target obtained by the hybrid algorithm are 14.29% and 60.53% higher than those obtained by the basic depth first backtracking algorithm respectively. Compared with weighted and single objective models, lexicographical multi-objective model improves the efficiency and quality of solution by 20.3% and 11.11%, respectively.