Study On Rules-based Architecture For Cognitive Decision Making

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.