关联网络结构下考虑随机和动态风险特征的协同生产调度及鲁棒优化

Aming to solve the problems of complex collaborative production relations existed among enterprises, uncertainty and risk interferences in outsourcing prodcution process, we construct multiple types of interdependent outsourcing production networks according to the production capability of outsourcing enterprises and analyze their characteristics. Based on such interdependent structure networks, we carry out the studies of collaborative production scheduling and robust optimization from randomness and dynamics perspectives: 1) Collaborative production decision and robust optimization with stochastic characteristic. By considering arrival probability of uncertain order at different interval time point, we construct four sub-models at the conditions of order arrival or not. In particular, two ordr merger decisions including initial merger and random merger of same type orders are considered. Furthermore, path constraints of order in interdependent outsourcing production networks are presented. Thus, the collaborative production decision model is constructed accordingly. Additionally, by taking into account arrival scenarios of multiple uncertain orders and their probabilities, the collaborative production stochastic expected cost model is constructed as well. 2) Collaborative production operation robust optimization with dynamic risk characteristic. By considering the dynamic variability and fragility of interdependent outsourcing production networks in risk environment, we decipt the failure propagation mechanism and design the disruption scenarios. In this way, a robust optimization model composed of expected cost and expected regret cost at different scenarios is constructed by using balance coefficient. The collaborative production issue of key component of intelligent equipment is finally stuided as the real instance to test our theory.

为解决外协生产过程中企业间存在的协同生产关系复杂、易受不确定和风险因素干扰问题，本项目依据外协企业具备的工序生产能力构建多类型关联外协生产网络并分析其性能。基于这种关联结构网络，从不确定性、动态性两个维度开展协同生产调度和鲁棒优化问题研究：1）考虑随机特征的协同生产决策与鲁棒优化。考虑不确定订单区间时间点到达概率，针对订单到达与否随机因素构建四类子模型，考虑同类订单初始和随机两种合并决策及在关联外协生产网络中的路径约束，依此构建协同生产决策模型。进一步地，考虑多不确定订单到达场景及实现概率，构建协同生产随机期望成本模型。2）考虑动态风险特征的协同生产运作鲁棒优化。考虑关联外协生产网络在风险环境下的动态可变性和脆性，描述风险动态传播机理并依此设计中断场景，采用平衡系数构建综合考虑不同场景集下期望成本与总体期望后悔成本的鲁棒优化模型。最后，以高端装备关键部件协同生产实例对理论研究成果进行验证。

高端装备产品关键部件在设计和生产上具有很高的复杂性，需要对其设计和生产工艺分解，并寻求外部设计与生产能力，通过网络化协同制造模式完成。为解决外协生产过程中企业间存在的协同生产关系复杂、易受不确定和风险因素干扰问题，本项目首先依据外协企业具备的工序生产能力构建多类型关联外协生产网络并分析其性能。基于这种关联结构网络，从不确定性和风险性两个维度开展协同生产调度和鲁棒优化问题研究，具体研究内容:1)协同网络结构下的协同生产调度优化。该部分内容考虑了四类协同生产网络（完整网络和风险环境下节点出现中断的非完整网络），构建最小化生产成本、等待成本、延期成本目标函数，并考虑了订单合并策略。由于模型的复杂性，设计了基于蚁群和蒙特卡洛算法结合的学习型算法，仿真结果表明了算法的稳定性和高效性。2)考虑随机特征的协同生产决策优化。考虑客户不确定订单区间时间点到达概率，针对订单到达与否随机因素构建四类子模型，考虑同类订单初始和随机两种合并决策及在关联外协生产网络中的路径约束，依此构建协同生产决策模型。进一步地，考虑多不确定订单到达场景及实现概率，构建协同生产随机期望成本模型，并转成确定性模型求解。3)考虑不确定特征下的协同生产鲁棒调度优化。考虑客户需求时间的不确定性，构建两阶段鲁棒优化模型，第一阶段在交货期未知情况下决策订单的指派和在协同企业处的生产排序，第二阶段在交货期实现后决策订单的开始生产时间。针对该模型，采用行列生成（C&CG）算法框架对模型求解。结果表明了两阶段鲁棒优化模型可显著改善调度策略的保守性。最后，以高端装备盾构机关键部件刀盘协同生产实例对理论研究成果进行验证，具有很好的应用价值。