Probabilistic Linguistic Petri Nets for Knowledge Representation and Acquisition With Dynamic Consensus Reaching Process

Fuzzy Petri nets (FPNs) are widely used in various fields for knowledge representation and reasoning. Nevertheless, the original FPN model has many limitations when applied in the practical situations. In response, this article aims to develop a new FPN model, named as probabilistic linguistic Petri nets (PLPNs), for representing and acquiring knowledge based on a dynamic consensus reaching process. First, to avoid losing the initial information, the probabilistic linguistic term sets (PLTSs) are adopted to represent the professional knowledge of domain experts. Subsequently, a knowledge acquisition approach is proposed to acquire the knowledge parameters of PLPNs. Then, a dynamic consensus reaching method is introduced during the knowledge acquisition process to increase the group consensus level among experts. Finally, a real-world risk assessment example concerning the subway fire system is presented to validate the usefulness and effectiveness of the proposed PLPNs. The results show that the new PLPNs are efficient and practical to represent and acquire expert knowledge with conflict and inconsistent opinions.