首页 > 硕士 > 工学 > 正文

基于PAC的神经网络PID控制的方法研究

Research of Methods of Neutral Network Controlling Based on PAC

作者: 专业:控制理论与控制工程 导师:江崇礼 年度:2010 学位:硕士  院校: 大连理工大学

Keywords

PID, Neutral Network, PAC, Self-tune, MATLAB

        目前工业自动化水平已经成为衡量其他各行各业现代化水平的一个重要标志,因此如何得到一种更为优越的控制方法成为了工业过程控制领域里一个重要的问题。PID控制是一种经典的控制方法,它以结构简单,不依赖于被控对象的数学模型,抗干扰能力强等优点,成为了过程控制领域一种重要的方法。本文基于PAC的基本构造以及工作原理,结合了三种典型的神经网络控制算法,利用BP神经网络PID控制和传统PID控制算法做比较,设计了基于三种算法的自整定PID控制平台。本文首先研究了经典PID控制原理,尤其是数字式PID控制器原理。分析了经典的PID参数整定方法。研究了神经网络的基本结构和原理,提出了几种经典的控制算法:BP神经网络控制,RBF神经网络控制以及CMAC神经网络控制。其次,本文重点研究了基于BP神经网络的PID控制原理,它的算法实现。针对工业过程控制具体的模型,做了基于BP神经网络的PID控制仿真图以及传统PID控制仿真图,对两者做了比较,证明了神经网络PID控制有着更加优越的控制性能。最后,本文研究了PAC的基本构造和工作原理,结合本文研究的BP神经网络PID控制,设计出了神经网络PID控制算法的模块,并得出了基于PAC的神经网络PID控制方法的流程图。
    The level of industrial automation has been an important symbol of weighing modern level of other area at the moment.PID controller, as a classical controlled methods, duing to its simple structure, not depending on the mathematical model of controlled object and the advantages of strong robustness, has been an important method of process controlling area. In this paper, BP neutral network PID control and traditional PID control were compared combined with three classical neutral network controlled methods. And a PID controlled platform was designed based on the basic structure of PAC and its operating principle.In this paper, classical PID principle of control was studied firstly, especially the principle of digital PID controller. Lately, classical PID parameter tuning methods was studied. It also studied the basic structure and principle. At last, several classical controlled algorithm were proposed, including BP neutral network control, RBF neutral network control and CMAC.Moreover, this paper put more emphasis on the research based on PID control principle and its algorithm realization. For the specific model in the industrial process control, this paper made PID controlling simulation figures and traditional PID control simulation figures. And the result of comparing both proved that neutral network PID control has better control performance.In the end, this paper studied the basic structure and work flow of PAC.A module was designed based on neutral network PID control combined with neutral network controlled algorithm. It also made the work flow of neutral network PID control based on PAC.
        

基于PAC的神经网络PID控制的方法研究

摘要4-5
Abstract5
1 绪论9-12
    1.1 课题的研究背景及意义9-10
    1.2 课题当今的研究现状10-11
    1.3 本文需要解决的关键问题11
    1.4 本文的结构组成11-12
2 PID控制器的基本原理12-18
    2.1 PID控制器12-14
        2.1.1 PID原理12-13
        2.1.2 PID各参数的作用13-14
    2.2 数字PID控制14-16
        2.2.1 位置式PID控制算法14-15
        2.2.2 增量式PID控制算法15
        2.2.3 两种算法的优缺点15-16
    2.3 PID参数整定16-17
    2.4 本章小结17-18
3 神经网络的基本原理18-30
    3.1 神经网络的模型结构18-20
    3.2 几种典型的学习规则20
        3.2.1 无监督的Hebb学习规则20
        3.2.2 有监督的Delta学习规则20
    3.3 几种典型的神经网络20-29
        3.3.1 BP神经网络21-24
        3.3.2 RBF神经网络24-27
        3.3.3 CMAC神经网络27-29
    3.5 本章小结29-30
4 神经网络PID控制基本原理以及应用30-48
    4.1 基于BP神经网络的PID控制30-45
        4.1.1 BP神经网络整定原理30-35
        4.1.2 MATLAB的背景和发展35-37
        4.1.3 常规PID控制系统37-38
        4.1.4 基于BP神经网络的PID控制系统38-45
    4.2 RBF神经网络和CMAC神经网络PID控制45-46
        4.2.1 RBF神经网络PID控制45-46
        4.2.2 CMAC神经网络PID控制46
    4.3 本章小节46-48
5 基于PAC的神经网路PID控制的方法研究48-53
    5.1 PAC简介48-51
        5.1.1 PAC的特征性能48-49
        5.1.2 PAC的工作原理49-51
    5.2 神经网络PID控制算法模块实现及流程图51
    5.3 基于PAC的神经网络PID控制流程图51-52
    5.4 本章小结52-53
结论53-54
参考文献54-56
攻读硕士学位期间发表学术论文情况56-57
致谢57-59
        下载全文需50


本文地址:

上一篇:拔头油火车装车控制系统的开发与研究
下一篇:长距离输油管线电动爬行器的研究

分享到: 分享基于PAC的神经网络PID控制的方法研究到腾讯微博           收藏
评论排行
公告