An Effective Way In Parameters Optimization Of Fuzzy Petri Nets

Fuzzy Petri Net is a very effective ways to solve the parallel processing capability of express knowledge and upgrade their knowledge, but the self-learning ability of the fuzzy system is defective. Fuzzy production rules in a number of parameters, such as weights, threshold and the degree of determination is heavily dependent on peoples'experience, it is difficult to obtain accurately and which preventing the processing of the fuzzy Petri Net knowledge and generalization ability. How to study the function of neural networks with the fuzzy Petri net and make the Fuzzy Petri Net through studying and training a number of samples. On the one hand, it has a certain generalization function. On the other hand, breaking away the experience dependence of building a FPN model of production rules of the parameters, so that the actual system parameters more in line with the situation. This article focus on studying the ability of the FPN problem.Based on analysis of FPN reasoning mechanism deeply, this paper introduces ant colony algorithm, and proposes a belt crossover and mutation factor fuzzy Petri Net parameter values of the effective means of optimization. For the specific problem of Knowledge Base System, the method and BP algorithm, genetic algorithm, the basic ant colony algorithm to carry out a comparative analysis. Simulation results show that with the use of crossover and mutation factor of ant colony algorithm to train the parameters of the correct rate of more accurate, high precision, to adapt to any complex FPN model, and the Fuzzy Petri Net has strong generalization ability and adaptive function.