物料供给不确定环境下的飞机移动生产线动态调度方法

Dynamic Scheduling Method for Aircraft Moving Assembly Line Under Uncertain Supply of Material

飞机装配所需的物料种类复杂且数量巨大,其准时供给往往存在较大的不确定性.为了有效解决物料供给不确定环境下的飞机移动生产线动态调度问题,将机器学习中的支持向量数据描述技术(SVDD)与传统的调度方法相结合,提出了基于SVDD的动态调度算法.通过软件CPLEX和元启发式算法求解不同物料供给延期情形下的调度模型,并将得到的优化结果作为样本对SVDD分类模型进行离线训练.在实时调度阶段,根据SVDD模型实现作业的提前、延期或准时执行的分类.基于该分类结果,利用局部前瞻搜索算法进一步对提前和延期作业的具体开始执行时间做出决策.数值实验结果证明了所提出的算法在响应速度和求解效果上均能满足实际飞机移动生产线动态调度的需求.

Due to the diversity and large quantity of material in aircraft assembly, there tend to be great uncertainties in just in time supply of material. To effectively solve the dynamic scheduling problem for aircraft moving assembly line with uncertain supply of material, the support vector data description (SVDD) in machine learning field is combined with the traditional scheduling method, and a dynamic scheduling approach based on SVDD is proposed. First, CPLEX and meta-heuristic are used to solve the mathematical model under different material supply delay conditions, and the optimized results are taken as the samples to train the SVDD classification model. In the real-time scheduling phase, the trained SVDD model is used to make the classified decisions on “advance”,“delay”, or “on schedule”. Based on the results of classification, a local look-ahead searching method is presented to make a further decision on specific starting time of jobs advanced or delayed. The computational results prove that the proposed algorithm can meet the actual requirement of dynamic scheduling for aircraft moving assembly line in both response speed and solution effect.