连铸机故障下炼钢-连铸动态调度混合协同进化算法

Hybrid Cooperative Coevolution Algorithm for Steelmaking-Continuous Casting Dynamic Scheduling with Continuous Caster Breakdown

连铸机故障问题不仅需要考虑设备指派和时间调整,而且涉及浇次计划的修复,是炼钢-连铸生产中一类复杂的扰动事件。本文针对该问题提出浇次计划修复策略,以最小化总等待时间和调度前后偏差的加权总惩罚为目标,建立了混合整数规划模型,结合问题联合决策及模型变量的特征,提出了遗传算法和粒子群算法相混合的协同进化算法。在算法中,设计了初始启发式修复策略和基于冲突炉次动态修剪解空间策略,提出了前向正推和反向逆推相结合的双向解码方式,并设计了混合协同进化算子,继而利用调整变量对钢材质量产生了不同程度的影响,设定了变量调整次序和基于柔性浇铸时间的可行性处理策略。最后,通过仿真实验验证了模型和算法的可行性和有效性。

The problem of continuous caster breakdown not only requires consideration of equipment allocation and time adjustment,but also involves repairing the casting schedule,which is a complex and large disturbance.To solve this problem,a modification strategy of the casting schedule was proposed.A mixed-integer programming model under this strategy is established.The objective was to minimize the total weighted penalty of the waiting time and scheduling deviation.Considering the characteristics of the joint decision and model variables of the problem,a cooperative co-evolution algorithm was designed by combining the genetic algorithm and the particle swarm optimization.First of all,the algorithm realized the dynamic pruning of the problem space through the initial heuristic strategy and the adjustment set based on the conflict charge.Then,the equipment assignment and time adjustment were coded independently.The bidirectional decoding combining forward and reverse inference was used to decode and evaluate.The evolutionary operators were designed based on the evolutionary mechanism to realize co-evolution.Furthermore,to ensure that the optimal solution of coevolution was feasible,a strategy based on flexible casting time was designed to repair and optimize the solution.Finally,the feasibility and effectiveness of the proposed model and algorithm were verified by simulation experiments on different scales.