CAJ | 학술논문

提出一种基于小生境自适应差分进化小波神经网络(NADE-WNN)的方法对不确定混沌系统进行控制.该方法利用小波神经网络学习未知模型混沌系统的动态特性并实施控制,为提高神经网络的学习精度和收敛速度.采用小生境自适应差分进化算法同时优化小波神经网络的结构和参数,简化网络结构,提高网络的学习精度和全局收敛性.仿真实验结果表明.在有外部干扰和参教摄动的情况下,NADE-WNN仍能对不确定混沌系统进行有效控制,且网络结构、控制精度和收敛速度都优于传统神经网络.
제출일충기우소생경자괄응차분진화소파신경망락(NADE-WNN)적방법대불학정혼돈계통진행공제.해방법이용소파신경망락학습미지모형혼돈계통적동태특성병실시공제,위제고신경망락적학습정도화수렴속도.채용소생경자괄응차분진화산법동시우화소파신경망락적결구화삼수,간화망락결구,제고망락적학습정도화전국수렴성.방진실험결과표명.재유외부간우화삼교섭동적정황하,NADE-WNN잉능대불학정혼돈계통진행유효공제,차망락결구、공제정도화수렴속도도우우전통신경망락.
A novel control method based on Niche Adaptive Differential Evolution Wavelet Neural Network (NADE-WNN) is proposed for uncertain chaotic system.The Wavelet Neural Network is used to study dynamic characters of uncertain chaotic system and control it.In order to raise learning accuracy and convergence rate,the structures and parameters of wavelet neural network are optimized by NADE algorithm at the same time in the model.The algorithm can get a best network structure and improve learning accuracy and the global convergence.The simulation results show that the NADE-WNN is still effective when there are external disturbance and parameter perturbation,and then the network structure,the control precision and convergence rate all outperform basic neural networks.