基于改进遗传算法的决策规则库自动维护与求精机制研究  

Research on Automatic Decision Rule Base Maintenance and Refinement Mechanism Based on Improved Genetic Algorithm

作　　者：魏红芹[1] 杨保安[1] 

机构地区：[1]东华大学旭日工商管理学院,上海200051

出　　处：《管理科学》2007年第5期69-75,共7页Journal of Management Science

基　　金：国家自然科学基金(70271002)

摘　　要：智能决策支持系统的性能和决策质量的优劣取决于知识库的内容和运行情况,随着系统和问题的日益复杂,知识库的规模越来越庞大,内容越来越复杂,需要提供有效的方法实现其优化管理。对智能决策支持系统中规则库的运行特性和可能的潜藏缺陷进行归纳分析,提出一种将传统优化方法和遗传算法相结合的二级规则库维护与求精机制,可以较好地识别和消除8种规则库缺陷,提高知识库的运行效率和推理求解效果。对各环节给出了具体的操作算法,可以实现在专家少量参与下的规则库自动优化。The performance of intelligent decision support system （IDSS） depends on the knowledge base quality. Knowledge base is becoming bigger and more complex, so it needs an effective method to optimize its management. A novel two-level rule base maintenance and refinement mechanism is proposed based on detail analyzing of the characteristics and potential detects of rule base used in IDSS. In the mechanism, conditional optimization approach and genetic algorithm are combined to recognize and eliminate eight kinds of detects in the rule base, which can improve the running efficiency and effects of knowledge base. Detail operations for each step are given and the automatic optimization of rule base can be realized with few help from experts.

关 键 词：智能决策支持系统 规则库求精 规则库维护 遗传算法 自动求精 

分 类 号：F224[经济管理—国民经济]