复杂系统的定性表示与推理  

The Qualitative Representation and Inference of the Complex Systems

作　　者：李波[1] 张世英[1] 李银惠[1] 

机构地区：[1]天津大学管理学院,天津300072

出　　处：《系统工程理论与实践》2002年第12期15-21,共7页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(69874028)

摘　　要：　基于人工智能中树图推理的思想,提出一种对复杂系统进行定性描述和推理的方法.该方法基于图论把一般的定性约束方程转化为一种约束图的表示形式,并在虚拟因果理论基础上,得到描述系统波动演化的各关键变量的约束影响图形式.最后,比较了Kuipers的QSIM和本文发展出的方法,表明采用图形化的方式来表征系统各变量之间所存在的约束关系,可以得到和原系统等价的但变量数大大减少的约束影响图形式.这种处理可以推动定性推理的研究从QSIM的实验室阶段走向实际应用领域.A method on the qualitative representation and inference of the complex systems has been proposed based on the thought of tree inference in artificial intelligence. With the help of graph theory, the general qualitative constraint equations have been transformed into the form of the constraint influence graph and the algorithm that can be used to choose the most important variables is obtained. Finally, the method proposed in this paper has been compared with QSIM by Kuipers, and the results show that the constraint influence graph can represent the relationship among the variables of the systems and prune many unimportant variables. This study can urge QSIM from the laboratorial stage to the real applications.

关 键 词：复杂系统 定性表示 推理 定性推理 人工智能 树图推理 

分 类 号：N941.4[自然科学总论—系统科学]