Abstract: To enhance the allocation of production resources and to enable coordinated and orderly production between steelmaking and continuous casting,based on the independent CVRP furnace planning and pouring planning models,a CVRP-based integrated steelmaking-continuous casting batch planning model was established with capacity limits and withdrawal penalties as the convergence constraints between steelmaking and continuous casting processes,and the model was solved using an improved genetic algorithm named Genetic Algorithm—Large Neighbourhood Search (GA-LNS) algorithm.The GA-LNS algorithm used genetic crossover,adaptive adjustment of variance probabilities,the idea of fusing LNS algorithm splitting and reconstruction and elite search to improve the search capability and solution quality of the algorithm.Finally,model validation and integrated batch planning were carried out using the CVRP algorithm and actual production data to verify the correctness of the model and the effectiveness of the algorithm,providing a new idea for enterprise batch planning.

Key words: steelmaking-continuous casting, batch planning, vehicle path problem, improved genetic algorithm

摘要： 为加强生产资源配置,使炼钢-连铸之间的生产协调有序,在独立的CVRP炉次计划和浇次计划模型基础上,将容量限制和撤销惩罚作为炼钢与连铸工序的衔接约束,建立了基于CVRP的炼钢-连铸一体化批量计划模型,并利用改进的遗传算法(GA-LNS)进行了模型的求解。GA-LNS算法采用遗传交叉、变异概率自适应调整、融合大邻域搜索算法(LNS)拆分与重构的思想以及精英搜索,提高了算法的搜索能力与解的质量。最后,采用CVRP算例和实际生产数据进行了模型验证和一体化批量计划编制,验证了模型的正确性与算法的有效性,为企业批量计划编制提供了新思路。

关键词: 炼钢连铸, 批量计划, 车辆路径问题, 改进遗传算法