Verification and validation of fuzzy rules-based human behavior models

To improve the fidelity and automation of simulation exercises, human behavior models have become key components in most military simulations. In this paper, we present a fuzzy Petri nets-based method for verification and validation of fuzzy rules-based human behavior models. This method consists of three parts: verification of fuzzy rule bases, static validation of human behavior models, and dynamic validation of human behavior models. We first use a formal description method to model human behavior models, then automatically map them to fuzzy Petri nets, and then generate reachability graphs to verify the incompleteness, inconsistency, circularity, and redundancy of the rule bases in human behavior models. We then construct a validation referent, and according to it search the fuzzy Petri nets to statically validate human behavior models. We finally reason the fuzzy Petri nets to dynamically validate human behavior models according to the validation referent.