Abstract: To solve the Steelmaking-Continuous Casting (SCC) scheduling problem with uncertain processing times,a directed graph-based flexible scheduling model was introduced,and an Estimation of Distribution Algorithm (EDA) was proposed.In the scheduling model,a directed graph model was used to describe flexible decisions and set time buffers for protecting the performance and feasibility of scheduling solution under uncertain environments.In the optimization algorithm,a stage-associated probabilistic model was developed to illustrate the relationship between decision variables,which could generate new high-quality populations.The stochastic simulation was used to evaluate the expected performance of flexible scheduling solutions.In the simulation,the Optimal Computing Budget Allocation (OCBA) technique was applied to allocate the limited computing resource for improving the evaluation efficiency of the population.The problem instances and computational results based on the practical production data demonstrated the effectiveness of the proposed flexible scheduling approach.

Key words: uncertain scheduling, steelmaking-continuous casting, buffer setting, estimation of distribution algorithm, optimal computing budget allocation

摘要： 针对加工时间存在不确定性的炼钢—连铸调度问题,构造了一种基于有向图网络的柔性调度模型,并提出一种基于分布估计的优化算法。在柔性调度模型中,使用有向图网络描述调度解的柔性决策,然后设置时间缓冲以保护不确定条件下调度解的性能指标和可行性;在优化算法中,使用基于工序关联的概率模型描述决策变量之间的关联关系以产生高质量的新种群,并采用随机仿真方法对柔性调度解的期望性能进行评价。在仿真过程中,运用最优计算量分配技术对有限计算资源进行动态分配以提高种群的评价效率。通过基于实际生产数据的问题实例和数值实验验证了所提柔性调度方法的有效性。

关键词: 不确定调度, 炼钢&mdash, 连铸, 缓冲设置, 分布估计算法, 最优计算量分配