基于量子进化算法的演化ANFIS模型及其应用

Evolution ANFIS Model and Its Application Based on Quantum Evolutionary Algorithm

利用量子进化算法对自适应模糊推理系统进行建模,从而利用自适应模糊推理系统和量子进化算法的两方面的优点来对种群结构进行优化,从而达到优化整个模糊推理系统的目的。通过仿真实例,将结合量子进化算法与自适应模糊推理系统分别应用于单输入单输出的模糊系统、多输入单输出模糊系统和多峰非线性模糊推理系统中,通过训练数据和测试数据得出性能的寻优跟踪路径及误差曲线进行比较。实验数据对比表明,ANFIS的缺点是精度低,GA-ANFIS的缺点是训练时间过长,而QEA-ANFIS主要摒弃了ANFIS训练的精确度上述两个系统的明显缺陷,既提高了精度又缩短了训练时间。

The quantum evolutionary algorithm（QEA） has received the widespread attention with its unique advantages such as fast convergence,short training time and global search capability.This paper mainly researches the quantum evolutionary algorithm by modeling with an adaptive network fuzzy inference system （ANFIS）,this method uses the two advantages of both quantum evolutionary algorithm and ANFIS to optimize the population structure.It can optimize the global fuzzy inference system.Through simulation examples,the model of QEA-ANFIS is applied to the single-input single-output fuzzy system,the multiple-inputs single-output fuzzy system and multimodal nonlinear fuzzy system.This method can get the performance of the optimization of path tracking through the training data and testing data,and also get performance of the error comparison.The simulation results show that ANFIS has low accuracy,and the GA-ANFIS has a long training time QEA-ANFIS overcome these two flaws and can improve the precision and the training time.