基于高斯微粒群优化的动态神经网络延迟系统辨识  

Dynamic neural network on Gaussian particle swarm optimization for delay system identification

作　　者：范剑超[1] 韩敏[1] 

机构地区：[1]大连理工大学电子信息与电气工程学院,辽宁大连116023

出　　处：《控制与决策》2010年第11期1703-1706,共4页Control and Decision

基　　金：国家自然科学基金项目(60674073);国家863计划项目(2007AA04Z158)

摘　　要：为提高神经网络对未知非线性大滞后动态系统的泛化能力,提出一种基于高斯微粒群优化的自适应动态前馈神经网络.在输入层与隐含层之间、隐含层与输出层之间分别加入动态延迟算子,可以高效地辨识出系统纯滞后时间,建立精确系统模型.此外,采用高斯函数和混沌映射方法平衡微粒群算法全局寻优能力,以克服提前收敛的缺陷,从而快速有效地自适应优化网络中的参数.仿真实验表明了该方法在非线性大滞后系统辨识中的有效性.In order to improve the generalization capacity of neural network for poorly known nonlinear dynamic system with long time-delay,a dynamic feedforward neural network on Gaussian particle swarm optimization algorithm is proposed. The dynamic delay operators are added between the input and hidden layer,output and last hidden layer,which can effectively identify the precise pure delay time.The nonlinear dynamic system model is built exactly to predict the system change.On the other hand,Gaussian function and chaos mapping are adopted to balance the global optimization ability to overcome premature convergence,which can optimize the parameters in the neural network structure adaptively.Compared with other methods in the experiments,the effectiveness of the proposed method on nonlinear long-delay system identification is shown.

关 键 词：动态神经网络 高斯微粒群 延迟系统辨识 预测 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]