【作者】 王莉；

【导师】 王明哲；

【作者基本信息】 华中科技大学 ， 控制理论与控制工程， 2009， 博士

【摘要】 系统的系统是由多个组件系统集结而成的集成系统，表现为人的认知决策过程的指挥控制系统是通过与其它组件系统的互操作活动而使得系统的系统成为有别于传统的烟囱式集成系统的管制型集成系统的关键组件系统。所以，基于组件系统紧密互动中的人的认知、判断和决策指挥行为研究应该成为集成系统体系结构研究中的重要内容，具有极其重要理论研究意义和工程应用价值。本文提出了一种认知决策框架，并通过建立该框架的基础性操作概念模型、指导性体系结构逻辑模型和动态性可执行模型，实现了认知决策体系结构的建模和分析研究。主要的研究和创新内容包括：层次化认知决策框架构成研究。分析了由识别认知和元认知模块构成的认知决策框架，通过识别认知对外界信息的处理和元认知对识别认知的监控呈现出不确定和时间紧迫条件下基于经验的认知决策行为过程；提出了认知决策框架层次化操作概念模型的建立流程。认知决策框架中各子模块的操作概念模型构建理论和方法研究。在环境感知模型中，采用基于概念格的关联规则挖掘方法实现指挥决策规则的提取；并将决策者对它们表现出的主观偏好信息和它们所固有的客观属性信息分别表示为语言判断矩阵和规范化的决策矩阵，通过对两类矩阵的操作，实现对规则的确认和排序。态势评估和计划制定模型中，提出了一种不确定规则的渐进式推理模式，建立了决策推理模型：首先采用D-S证据理论对定性或不确定表述的规则进行初步处理，得到初步合成结果，然后将其用于指导建立定量化描述的贝叶斯网推理模型。由于元认知对识别认知的监控是一个反复定性化循环判断的动态过程，采用模糊认知图的因果表述及其逐步动态推进式的表示形式来体现这个过程中的“循环”和“动态”两大特性，建立元认知动态监控模型。认知决策体系结构逻辑模型和可执行模型构建的理论和方法研究。首先，以对表现为理论模型的认知决策操作概念模型为基础，采用系统的系统体系结构指导性文件DoDAF为依据，通过构建相关视图产品，建立认知决策框架的功能体系结构逻辑模型。然后，根据建立的体系结构逻辑模型，结合操作概念模型中的动态描述，采用着色Petri网建模工具建立对应的动态可执行模型。最后，对基于规则的认知决策体系结构可执行模型进行静态结构和动态行为的分析，完成模型的逻辑正确性检验；通过模拟指挥控制组件系统与其它组件系统间互动的认知决策可执行模型的仿真分析，实现并证实认知决策框架在整体系统中期望的指挥控制能力。更多还原

【Abstract】 Based on its interaction with the other component systems in directed Systems of Systems, Command and Control component system which is represented by the human cognitive decision making process is the key component to make System of Systems different to the stove-piped systems of the past. The behavioral characteristic study of human cognition, judgment and decision making for the integrated system developed with the interaction should be the most important content of the studies for integrated cognitive architecture, and it will present huge significance of theory research and value of engineering application.In terms of the cognitive decision making framework presented in the paper, the study on cognitive architecture modeling and analysis for the framework are realized by the creating of the basic operational concept models, the guiding logic architecture models and the dynamic executable model. Contributions of the dissertation are as follows:Firstly, the study of cognitive decision making framework is hierarchically proposed. Undering the conditions of uncertainty and time stress, human cognitive decision making processes based on their experience are represented by a cognitive decision making framework, which are reflected by the recognitive processing for environment information and matacognitive skills for monitoring and regulating the processing. A hierarchical modeling process of the operational concept for the cognitive decision making framework are presented.Secondly, the modeling theories and approaches to the operational concept models of modules of cognitive decision making framework are proposed one by one. In the concept models of the process of environment cognition, at first, the decision-making rules are extracted based on the concept lattice theory with association rules mining. And then, the subjective preference information of decision maker and the objective information of attributes about those extracted rules are respectively represented by the language judgment matrix and the decision making matrix to rank and confirm the final useful rules set. In the process of situation assessment and plan framework, a gradual inference approach of decision making grounded on the uncertain rules is proposed, and an inference model for decision making is created too: Firstly, the qualitative and uncertain expression accompanied with rules can be disposed based on the D-S theory, and the elementary consequences of instruction can be gained. Secondly, combining the consequences of instruction with the experts’ experience, the quantitative Bayesian Nets model can be created reasonably and perfectly. Moreover, since the monitoring process of metacognition on recognition is a qualitatively circular judgment process, the dynamic fuzzy cognitive map is used to describe the characteristics of "circle" and "dynamic" of such process and create the model for metacognition by its causal representation and the dynamic pushing expression form.Finally, the modeling theories and methods of logic architecture models and executable models for the cognitive decision making framework are proposed. Based on the created operational concept models of the modules in the cognitive framework, the guidance documents of modeling for integrated architecture, i.e., DoDAF is adopted to create the logic model of the cognitive functional architecture. With the combination of the created logic model and the dynamic description of the operational concept models, a dynamic executable architecture model is created by the Colored Petri Nets modeling tools. A set of static structure and dynamic behavior analysis are implemented to verify and evaluate the executable model. Grounded on the simulational analysis for the model of cognitive decision making which represents the interaction between the Command and Control component system and the other component systems, the expectant ability to command and control in the cognitive decision making process is achieved and confirmed totally.更多还原