基于交互仿真的生产决策专家系统构建方法  

Constructing method of production decision making expert system based on interactive simulation

作　　者：张晓冬[1,2] 张志强[2] 陈进[2] 段爽月[2] 

机构地区：[1]北京科技大学东凌经济管理学院,北京100083 [2]重庆大学机械传动国家重点实验室,重庆400030

出　　处：《计算机集成制造系统》2013年第2期404-410,共7页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金项目(70971146);高等学校博士学科点专项科研基金项目(20120006110034);中央高校基本科研业务费项目(FRF-SD-13-004B)~~

摘　　要：为提高生产决策水平,提出了基于人机交互仿真的生产决策专家系统构建方法。针对生产决策专家知识难以采集和控制策略复杂等特点,采用神经网络和专家遴选算法实现对专家知识的评估和优选,结合融合分类算法解决知识推理的问题,并提出TR-TREE算法构建专家系统规则解释机制。以某摩托车发动机关键零部件生产单元的换线决策为应用背景,对所提出的实现原理和关键技术进行了应用实践。与基于经验的人工决策相比,基于专家系统的换线决策对生产目标的优化作用显著,并具有良好的学习机制。To improve the production decision making level in production system, the constructing method of production decision making expert systems based on human-machine interactive simulation was proposed. Aiming at the difficulties of knowledge collection and controlling strategy complexity, technologies of interactive simulation and expert knowledge selecting were developed. The knowledge reasoning was realized by combining with fusion classifica- tion algorithm, and the explanation mechanism was constructed by TR-TREE algorithm. Taking the line switching decision in a motor engine production cell as an example, the method was practiced and validated. The case study showed that the expert system could effectively optimize the production objectives with high learning ability.

关 键 词：生产决策 专家系统 交互仿真 摩托车发动机 

分 类 号：TH11[机械工程—机械设计及理论]