基于支持向量机的非线性系统辨识及控制研究

发布时间：2018-01-06 10:27

本文关键词：基于支持向量机的非线性系统辨识及控制研究　出处：《河南师范大学》2016年硕士论文　论文类型：学位论文

【摘要】：现实生活中大部分过程都是非线性的,传统的线性辨识方法难以辨识非线性系统,进而难以实施有效内模控制。针对上述问题,本文设计了一种基于SVR的非线性系统辨识及控制策略,主要内容如下:首先,对非线性系统辨识和控制进行深入的分析研究,提出了使用SVR来进行非线性系统辨识及控制研究。为了自动获取SVR的相关参数,提出了基于模糊差分进化(FDE)算法来自动获取SVR相关参数。其次,设计了基于FDE-SVR的非线性系统辨识方法。为了说明FDE算法的优越性,进行两次试验:第一次,使用模糊差分进化算法自动获取相关参数,然后使用获取到的参数进行非线性系统辨识;第二次,使用标准差分进化算法自动获取相关参数,然后用获取到的参数进行非线性系统辨识。实验结果表明:基于FDE-SVR的非线性系统辨识方法能够提高辨识精度和辨识速度。最后,基于FDE-SVR实现对非线性系统内模控制。首先,使用FDE-SVR辨识出内模控制中的内部模型M;然后,使用FDE-SVR辨识出内模控制中的逆模型C;最后,把辨识得到的内部模型M和逆模型C进行组合,得到基于FDE-SVR内模控制系统。通过使用阶跃信号、正弦信号、方波信号和锯齿信号对基于FDE-SVR的内模控制系统进行跟踪检验,结果表明该控制系统能够对给定的输入信号实现有效的跟踪,且具有较强的鲁棒性和抗干扰能力。
[Abstract]:In real life, most processes are nonlinear, the traditional linear identification method is difficult to identify nonlinear systems, and then implement effective internal model control. In this paper, a nonlinear system identification and control strategy based on SVR is designed. The main contents are as follows: firstly, the identification and control of nonlinear system are deeply studied. In order to obtain the relevant parameters of SVR automatically, a nonlinear system identification and control study based on SVR is proposed. A fuzzy differential evolution (FDE) algorithm is proposed to obtain SVR parameters automatically. Secondly. A nonlinear system identification method based on FDE-SVR is designed. In order to illustrate the superiority of FDE algorithm, two experiments are carried out: first, fuzzy differential evolution algorithm is used to automatically obtain the relevant parameters. Then the obtained parameters are used to identify the nonlinear system. The second time, the standard differential evolution algorithm is used to automatically obtain the relevant parameters. The experimental results show that the nonlinear system identification method based on FDE-SVR can improve the identification accuracy and speed. The internal model control of nonlinear system is realized based on FDE-SVR. Firstly, FDE-SVR is used to identify the internal model of internal model control. Then, the inverse model C in internal model control is identified by using FDE-SVR. Finally, the identified internal model M and inverse model C are combined to obtain the internal model control system based on FDE-SVR. The square wave signal and sawtooth signal are used to track and test the internal model control system based on FDE-SVR. The results show that the control system can effectively track the given input signal. And it has strong robustness and anti-interference ability.
【学位授予单位】：河南师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP18;N945.14

【相似文献】