中文摘要：

在工件加工时间具有恶化效应的单机环境下, 研究初始计划执行中计划外多个新工件到达的干扰管理问题. 将加工成本作为初始目标, 将工件相对于初始完工时间的延迟作为扰动目标, 构建多目标干扰管理模型. 结合归档式多目标模拟退火算法在全局寻优方面的优势, 与非支配排序遗传算法在快速收敛到Pareto有效前沿的局部搜索优势, 设计了混合元启发式算法在全局搜索和局部搜索之间进行平衡. 通过分析问题Pareto最优解特性, 可以进一步有效降低混合元启发式算法的搜索空间, 提高收敛速度和输出有效前沿的质量. 最后, 通过随机生成算例进行数值实验, 验证混合算法对求解干扰管理问题的有效性和Pareto最优解特性对于算法性能的改进.

英文摘要：

In single machine scheduling with deteriorating processing time, we study the problem of dealing with the arrival of multiple unexpected orders. We build up the bi-objective model where original objective is based on system operational cost, while the deviation objective is based on the delay of job＇s completion time with respect to its original completion time. In order to effectively solve the model, we combine simulated annealing-based multi-objective optimization algorithm, which is good at jumping out of local optimality, with non-dominated sorting genetic algorithm, which is good at fast converging to Pareto front. And we design a hybrid algorithm to balance between exploration and exploitation. By analyzing the Pareto optimal property, we could further effectively narrow the searching space of hybrid algorithm, speeding up convergence and improving Pareto front quality. Finally, by randomly generating and solving numerical problem instances, we show that our hybrid algorithm is effective for the disruption management problem, and Pareto optimal property could significantly improve the performance of hybrid algorithm.