Verification of Human Decision Models in Military Simulations

Abstract

Human decision models, often represented as rules or predicates, are key components in most military simulations. Although there exist some verification techniques for rule bases, they are not enough to assure the correctness of human decision models. Based on an analysis of the characteristics of human decision models in military simulations, this paper presents a method for verifying human decision models. A formal description method is developed to describe human decision models, which are automatically mapped to fuzzy Petri nets. A set of formal verification criteria for human decision models are developed, based on which verification is grouped into two classes: weak verification and strong verification. By searching the fuzzy Petri nets, weak verification is implemented, and by reasoning the fuzzy Petri nets, strong verification is implemented