Fuzzy Neural Petri Nets for Expert Systems

Fuzzy Petri Nets (FPN) is a powerful modeling tool for knowledge-based systems based on fuzzy production rules. But the lack of learning mechanism is the weakness of fuzzy systems. In this paper, a expert system modeling method called Fuzzy Neural Petri Nets (FNPN) is proposed. This model has both the features of a fuzzy Petri net and learning ability of a neural network. Being trained, a FNPN model can be used for dynamic knowledge representation and inference. The back propagation algorithm of neural networks is introduced into FPN. And the parameters of fuzzy production rules in FNPN can be learned and trained by this means. At the same time, different layers can be learned and trained independently. An example is included as an illustration. It is proved in this paper that the FNPN has powerful reasoning ability and adaptation ability. At the same time it can be regarded as a conceptual and practical artificial intelligence tool for the expert system.