Abstract: Aiming at the characteristics of multi-buffer and multi-constraint within the production of steelmaking-continuous casting,a batch-based scheduling approach comprised heuristic rules and memetic algorithm was proposed.By using unit-specific event-point continuous-time representation,a mixed integer linear programming model was built and thus three types of buffers were refined which included finite capacity,infinite capacity,and infinite stocking after processing.Two heuristic rules were proposed,the in-batch was utilized for ensuring the smoothness of production by allocation of these buffers,and the between-batch was employed for satisfying the setup constraint between two successive batches by timing.The memetic algorithm was designed,in which the in-batch and between-batch rules were integrated in the initialization and decoding process respectively and a composite matrix-based variable neighborhood search method was exploited for local optimization.Experimental studies of one given case and several randomly generated instances demonstrated the effectiveness and competence in local exploration and global optimization.

Key words: production scheduling, steelmaking continuous casting, heuristic rules, memetic algorithm, variable neighborhood search

摘要： 针对炼钢连铸生产多缓冲多约束的特点,基于浇次调度提出一种混合启发式规则与文化基因算法的调度方法。建立了炼钢连铸生产调度模型,提炼出有限、无限和可加工三类缓冲区。提出浇次内和浇次间两种启发式规则,通过缓冲区分配和定时来保证同一浇次内的生产顺行和各浇次间的准备时间约束。设计了文化基因算法,将启发式规则融进初始化和解码过程,构造基于优先组合矩阵的变邻域搜索算子改进其局部优化。通过某炼钢厂的实例及随机实验表明所提方法有效,且具有优异的局部拓展与全局优化能力。

关键词: 生产调度, 炼钢连铸, 启发式规则, 文化基因算法, 变邻域搜索