基于极限学习机的非线性内模控制  

Nonlinear Internal Model Control System Based on Weighted Regularized Extreme Learning Machine

作　　者：唐贤伦[1] 周家林[1] 张娜[1] 刘庆[1] 

机构地区：[1]重庆邮电大学工业物联网与网络化控制教育部重点实验室,重庆南岸区400065

出　　处：《电子科技大学学报》2016年第1期96-101,共6页Journal of University of Electronic Science and Technology of China

基　　金：国家自然科学基金(60905066)

摘　　要：针对非线性的内模控制的逆模难以求解的问题,该文提出一种基于改进极限学习机(MELM)的非线性内模控制方法。在基本的极限学习机模型中加入L1和L2范数罚函数,然后将改进极限学习机算法用于建立非线性系统的内模型和逆模型。仿真实验中,选取4种典型信号进行跟踪,并检验了系统的抗干扰能力和系统参数发生变化时的鲁棒性,通过将MELM和最小二乘支持向量机(SVM)以及极限学习机算法进行对比,表明基于MELM的内模控制方法对非线性系统具有更好的控制性能、较强的抗干扰能力和鲁棒性能。A nonlinear internal model control system based on the modified extreme learning machine algorithm is proposed. L1 norm penalty and L2 norm penalty are used to modify the model of ordinary extreme learning machine, then the modified extreme learning machine （MELM） is employed to establish the internal model and inverse model of nonlinear systems. In the simulation experiment, four typical signals are selected for tracking, and then the system＇s anti-interference ability＇ and the robustness when the system parameters change are verified. The results show that, compared with least-squared support vector machine and extreme learning machine algorithm, the proposed internal model control has better control performance, anti-interference ability and robustness for nonlinear system.

关 键 词：极限学习机 内模控制 L1范数罚函数 L2范数罚函数 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]