连铸机开浇炉次与时间决策的多目标优化

摘要

For the integrated decision making of charge selection and sequencing in the batch plan as well as the casting start time for continuous casters,a multi-objective optimization model was developed.The objectives were set to minimize the penalty of production batch plan implementation in steel plans,the amount of metal stocked in production line and the non-effective usage amount of high quality hot metal.According to the characteristics of model,an improved solving multi-objective evolutionary algorithm was derived based on Non-dominated Sorting Genetic Algorithm-Ⅱ (NSGA-Ⅱ).The chromosome was represented with the serial number of all charges in the charge batching pool to decrease the invalid searching space of solution.To reduce the computational complexity,the computation order of elitist solution in classical NSGA-Ⅱ was modified and the crowding distance calculation times for individuals were restricted.A fuzzy selection method was employed to choose the final solution among Pareto solutions.Computational tests on the real operation data of a steel plant showed that the proposed multi-objective optimization model was conducive to the stable control of charge's casting cycle on continuous casters,and the modified algorithm was better than the classical NSGA Ⅱ and Strength Pareto Evolutionary Algorithm-Ⅱ (SPEA-Ⅱ).%针对连铸机开浇决策中的炉次选择、排序与开浇时间确定的多目标优化问题,以炼钢厂生产批量计划执行情况的总惩罚、生产线积压金属量、优质铁水非有效利用量最小为目标函数,构建了连铸机开浇炉次与时间决策的多目标优化模型.针对该模型特点设计了改进的非支配排序遗传算法,以预选池内选择的炉次序号为基因的编码方式减小模型解的无效搜索空间,采取调整传统精英解集的计算顺序、限定计算拥挤距离个体数目的改进措施来减轻计算负荷,利用对Pareto解进行模糊选优的方法确定最终优化解.以某钢厂的生产实例数据测试表明,该模型有利于连铸生产炉次浇铸周期的稳定控制,算法效率优于传统的非支配排序遗传算法和强度Pareto进化算法.