数据驱动的Job Shop生产调度知识挖掘及优化  

Data-driven Job Shop production scheduling knowledge mining and optimization

作　　者：王家海[1] 陈煜[1] 

机构地区：[1]同济大学机械与能源工程学院,上海201804

出　　处：《计算机工程与应用》2018年第1期264-270,共7页Computer Engineering and Applications

摘　　要：在车间作业调度问题的求解过程中,从调度数据中挖掘调度知识,指导优化,对调度方案的精确求解至关重要。因此在分析调度领域海量数据和专业知识的基础上,建立基于本体的调度知识关系模型及知识表示;考虑数据挖掘与知识结合的关系,集成支持向量机和CART决策树学习算法,以实现数据驱动的调度规则获取,并分析调度知识挖掘框架下调度规则的挖掘过程;将调度知识和人工鱼群算法相结合用于生产调度的优化计算,改进人工鱼初始化过程。设计对比实验,实例验证表明算法效率获得较大提高,能够获得更接近实际情况的优化方案,有效减少作业总通过时间,提高了生产效率。During the evaluation process of the Job Shop scheduling problem, the knowledge hidden in mass data of the plant plays the vital role in conducting the production optimization. Based on the analysis of the mass data and specialized knowledge in scheduling domain, ontology-based relation model is established and the knowledge is represented. Considering the importance of the combination within the data mining and knowledge, the decision tree CART(Classification and Regression Tree)algorithm integrated with SVM(Support Vector Machine)is introduced to obtain the data-driven scheduling rule, an instance is laid out to show the process of rule acquisition under the scheduling knowledge mining framework. When it comes to the implementation, the knowledge combining with artificial fish swarm algorithm is applied to artificial fish initialization optimum design. Finally, a calculation analysis of samples is carried out and the results show the effectiveness and accuracy of the improved algorithm, with more practical solution and effectively reduce the makespan and enhance productivity.

关 键 词：数据驱动 本体 知识挖掘 生产调度 决策树规则 人工鱼群算法 

分 类 号：TB497[一般工业技术]