复杂信息系统分布式决策融合模型及应用研究  

Research of Distributed Decision-Level Fusion Model of Complex Information System with Application

作　　者：魏守智[1] 赵海[1] 王刚[1] 张晓丹[1] 

机构地区：[1]东北大学信息科学与工程学院,沈阳110004

出　　处：《计算机科学》2005年第4期22-23,30,共3页Computer Science

基　　金：国家自然科学基金(69873007);大庆油田科技攻关项目的资助

摘　　要：针对复杂信息系统信源为涉及多个不同专业背景的多源异类、异构信息,融合过程必须将多个异构模型集成起来进行联合问题求解的特点,提出了一种新的复杂信息系统分布式层次化决策融合模型,实现了集成神经网络组与专家系统及改进的分布式D-S证据推理模型的融合、特定领域中理论及经验模型与上述模型的融合,改进的D-S证据推理模型解决了当证据矛盾时导致错误结果的问题。油田剩余油分布及潜力预测的应用,表明提高了预测的精度和可靠性,为复杂融合系统的工程实现提供了重要的提示。Aiming at the characteristic of complex information system, whose information source deal with multi- source and different class as well as different structure information in multiple different specialty, whose fusion pro- cess must integrate multiple models and carry through combine problem-solving, a new distributed and hiberarchy de- cision-level fusion model of complex information system is proposed. Integrated neural networks groups models fuse expert system and improved D-S evidence reasoning model in distributed and hiberarchy system architecture, domain theory models and experiential models fuse above models are implemented. Improved D-S evidence reasoning model solve the problem that evidence combine produce false results when evidences are paradoxical. Oil-field remaining oil distribution and potential forecast system has applied above models, it shows that the precision and reliability of fore- cast are improved, this paper provides an important guidance to engineer implementation of complex information fusion system.

关 键 词：复杂信息系统 融合模型 分布式决策 应用 D-S证据推理 证据推理模型 集成神经网络 剩余油分布 专业背景 异构信息 问题求解 模型集成 专家系统 经验模型 特定领域 潜力预测 工程实现 融合系统 层次化 模型解 可靠性 

分 类 号：TP315[自动化与计算机技术—计算机软件与理论] TP274.2[自动化与计算机技术—计算机科学与技术]