一种基于类覆盖和粒子群优化的模糊神经网络系统

提出一种基于类覆盖获取有向图和粒子群优化方法的模糊神经网络模式识别系统模型，该模型利用改进的贪心算法获得半径较均匀的超球体类覆盖，再利用超球体类覆盖实现模糊输入空间划分和模糊IF-THEN规则提取，以此实现模糊神经网络系统的结构辨识；采用改进的模糊加权型Mamdani推理法确定系统的输出，并使用基于粒子群优化的算法对系统参数进行精炼，使系统具有很好的强壮性和识别率．对11种矿泉水味觉信号的识别实验结果证明了该系统的可行性和有效性．

A Fuzzy Neural Network System Based on the Class Cover and the Particle Swarm Optimization

A fuzzy neural network identification model is developed based on the class cover catch digraph and the particle swarm optimazition method. As structure learning of the fuzzy neural network,an improved greedy algorithm is presented for getting the hypersphere class covers with relatively even radii,which is used to partition fuzzy input space and extract fuzzy IF-THEN rules. A weighted Mamdani inference mechanism is improved for the system output and a particle swarm optimization(PSO)-based algorithm is used for optimizing system parameters that improve the system's correct classification percentages and robustness. Experimental results show that the system is feasible and effective to identify 11 kinds of mineral waters by its taste signals.