棒线材轧制批量调度多目标混合优化

Multi-objective Hybrid Optimization of Lot Scheduling for Bar Mill Process

在分析批量调度问题特征的基础上建立了精轧工序轧制批量调度的数学模型,采用混合自适应多目标进化算法进行求解.在该算法中,采用全局搜索与局部优化相结合来加快算法的收敛速度,基因修正与罚函数相结合来解决约束问题,运用免疫共享方法维护种群的多源性,根据评估结果自适应改变遗传操作的概率.应用生产实际数据进行测试,表明该调度方法能获得所需的Pareto优化前沿.

Based on the analysis of the management characteristics of the bar mill process, the model for the scheduling problem is set up. The optimal solutions are obtained by using hybrid adaptive multi-objective evolutionary algorithm （HAMOEA）. In HAMOEA, the convergence speed of the algorithm is accelerated by combining global search with local search. The constrain problems are resolved by combining the genes modification with the punishment function method. The diversification of population is maintained by using immune share method. The probabilities of genetic operations are adjusted adaptively according to the evaluation of operation during searching. The approach was tested using the real production data, and the results show that the Pareto front can be obtained efficiently.