Target Threat Assessment using Rule-Based Joint Fuzzy and Probabilistic Networks

Threat assessment is one of the most important pillars of data fusion systems. In this paper, we use two graphical models: fuzzy cognitive map and bayesian network to implement a complete threat assessment network. The structure of this network includes numerous variables of threat assessment and relates them well to each other. Given the uncertainty in all threat assessment issues, various types of uncertainty and how to deal with them are considered in this article. A comprehensive review has also been carried out on a variety of methods for incorporating both types of fuzzy and probabilistic uncertainties and a new approach is proposed. In this method, two separated fuzzy and bayesian networks are used to consider uncertainties. The approach of the proposed method is fully described, step-by-step. Furthermore, this paper addresses the major challenges of the threat assessment problem and shows that the proposed method is capable of solving these issues. To illustrate the effectiveness of the proposed method, a set of qualitative and         quantitative validation criteria is presented. As a test a scenario for air targets is simulated and the results of the proposed method are qualitatively and quantitatively compared with fuzzy cognitive map and    bayesian network methods. These results indicate that the proposed method works better than other      methods regarding root mean square error, total and trivial sensitivity degree and seperation degree. Moreover, the effectiveness of the proposed structure and method has been confirmed by experts in the field of battle management.