Research On Reasoning Under Uncertainty Methods In Situation Assessment

In information war, the situation assessment focusing on cognitive activities such as awareness, comprehension and prediction of battlefield situation has become one of the central technologies in modern command decision. The situation assessment is essentially a reasoning process on decision level, which tries to sense accurately and thoroughly according to the present situation, gradually recognizes adversarial intent and deployment, and helps our commanders make decision. Therefore, as a main factor impacting on the quality of decision, adversarial intent recognition is becoming a hot and difficult issue in situation assessment.Limited by technological conditions, the early adversarial intent recognition is almost men's work. Since the war form has been revolutionized, the processing of abundant data and information in the complex battlespace is beyond men's ability. A reasoning method which can accurately, quickly and automatically recognize adversarial intent is badly needed to increase efficiency and effectiveness in decision. Due to uncertainty, incompleteness and inaccuracy in battlespace, the key point of adversarial intent recognition in situation assessment is that how to make reasonable and quasi reasonable conclusion from uncertain data and knowledge.1 Aiming at the uncertainty in situation assessment, the paper proposes a two-stage reasoning model of adversarial intent recognition based on situation comprehension, which divides adversarial intent recognition into two stages, intention prediction and intention reference. Determining the confidence of all the possible adversarial intent through stimulation of adversarial decision process, and then inferring the real intention with the observed adversarial actions, it decreases the complexity in reasoning model of adversarial intent recognition.2 Aiming at the adversarial faith, their comprehension of ours, the higher level intention of adversarial operational unit and that of the corresponding level, along with the interrelationship among these factors, the paper employs the methods of uncertainty causal reasoning based on fuzzy Petri net to make intention prediction, presents the confidence of all the possible adversarial intent, and offers reasons for inferring the real adversarial intention according to the actions.3 Aiming at the intention of adversarial operational unit at the corresponding level and the actions which may be taken to fulfill the intention, it presents the potential relationship among the factors with Bayesian Networks, and proposes an intention reference stage modeling methods using fuzzy numbers to reduce the difficulty of gaining prior probability. The result of synthesized intention prediction and all the proofs of adversarial actions from sensors and others data resources are used to infer the real intention. 4 By analyzing a specific military scenario, the paper proves the feasibility and validity of the proposed model based on situation comprehension, and proposes a new way to computer realization of abundant multi-source information synthesis assisting the commanders in the process of adversarial intent recognition and judgment.