基于遗传算法的变结构神经Petri网模型

A Neural Petri Net Model with Variable Topology Based on the Generic Algorithm

将Petri网理论引入到神经网络模型的研究中，依据生物学神经系统的可塑性，通过适当扩展规范Petri网，建立了人工神经元状态转化的Petri网模型．在此基础上给出了利用遗传算法设计变结构神经Petri网的方法，该方法不仅可以得到满足要求的网络参数，而且能对网络的拓扑结构进行优化．仿真结果验证了该方法的可行性．

The Petri net theory is used to describe the models of artificial neural networks. Based on the plasticity of biological neural networks, a Petri net model for artificial neuron status transformation is presented by a proper extension of the standard Petri net. By taking advantage of the genetic algorithm, a method of constructing the neural Petri net with variable topology is proposed. Not only can the parameters of networks that meet the requirements be obtained, but also the optimization of the topological structure of the networks can be carried out. The feasibility of the method is verified by simulation results.