A Model for Dynamic Knowledge Representation and Learning

Knowledge of complex systems is often imprecise and subject to frequent modifications. Consequently, creating a knowledge representation and inference model that can adapt to changes in information is crucial. In this paper, we integrate the functional link into the fuzzy Petri net and propose a generalized model called functional link fuzzy Petri net (FLFPN). This model retains the explanatory ability of fuzzy Petri nets while acquiring the powerful learning ability of functional link neural networks. Finally, since the forming of congestion is highly sensitive to traffic situations in peak hours, we use FLFPN to predict traffic situations of an expressway ramp, which shows a significant improvement in the prediction accuracy, as compared with the traditional FPN model.