离散型动态规划模型的知识表示及其IBFS算法研究  

A Research on the Knowledge Representation for Discrete Dynamic Programming Model and Its IBFS Algorithm

作　　者：胡祥培[1] 钱国明[1] 胡运权[1] 

机构地区：[1]哈尔滨工业大学管理学院运筹学教研室

出　　处：《哈尔滨工业大学学报》1996年第3期119-126,共8页Journal of Harbin Institute of Technology

基　　金：国家自然科学基金

摘　　要：针对运筹学模型表示中存在的问题，本文以离散型动态规划模型为研究对象，从解决动态模型的知识表示与基于知识的推理机制入手，通过剖析动态规划问题的决策过程及其状态演变特征，运用人工智能状态空间理论，将动态规划模型的求解转化为状态空间图中最佳路径的搜索，提出了以六元组Ｍ＝（Ｉ，Ｇ，Ｏ，Ｔ，Ｄ，Ｓ）表示离散型动态规划模型的知识表示方法－ＩＧＯＴＤＳ表示法；并引入最优评价函数，研究了基于状态空间图的求解离散型动态规划模型ＩＢＦＳ搜索算法．本文的工作有利于促进并深化运筹学应用的知识化、智能化研究．According to the problems which exist in the representation of Operations Research(OR) model, this paper focuses on Discrete Dynamic Programming Model (DDPM) to solve such problems as knowledge representation for dynamic model and its knowledge-based inference mechanism. Having analysed the decision procedure of a dynamic programming problem, we know that the solution procedure of a Dynamic Programming Model can be transformed into a searching operation for the optimal path in its state-space diagram. Applying the state-space theory of Artificial Intelligence(AI), a knowledge representation for DDPM, named IGOTDS method which uses sextuple M=(I, G, O, T, D, S) to describe the model, is presented in this paper.The IBFS search algorithm to solve DDPM based on the state transition diagram has been studied by means of making optimal evaluation functions. This research is beneficial for advancing and deepening the research in the development of knowledgeablization & intellectualization for the applications of OR.

关 键 词：动态规划模型 知识表示 人工智能 IBFS算法 

分 类 号：O221.3[理学—运筹学与控制论]