文档介绍：摘要
经典PID控制算法作为一般工业过程控制方法应用范围相当广泛,原则上讲它并不依赖于被控对象的具体数学模型,但算法参数的整定却是一件很困难的工作,更为重要的是即使参数整定完成,由于参数不具有自适应能力,因环境的变化,PID控制对系统偏差的响应变差,参数需重新整定。针对上述问题,人们一直采用模糊、神经网络等各种调整PID参数的自适应方法,力图克服这一难题。一般情况下,一个自适应控制系统能够运行,其相应的参数要适应现场状况的变化,因此就必须根据现场的数据对相应的参数进行在线辨识或估计。对非时变参数可以通过一段时间的在线辨识确定下来,但对时变参数系统,必须将这个过程不断进行下去,因此要求辨识速度快或参数变化速度相对较慢,极大地限制了自适应技术的应用。为克服这种限制,本文利用文献[1]的思想,将神经网络的技术应用于参数辨识过程,结合经典的PID控制算法,形成一种基于BP神经网络的自适应PID控制算法。这一算法的本质是应用神经网络建立系统参数模型,将时变参数系统的参数变化规律转化为神经网络参数模型,反映了参数随状态而变的规律,即当系统变化后,可直接由模型得到系统的时变参数,而无需辨识过程。在神经网络参数模型的基础上,结合文献[1]已知系统模型下PID控制参数的计算,推导出一种自适应PID控制算法。通过在计算机上对线性和非线性系统仿真,结果表明了这种自适应PID控制算法的有效性。
关键词自适应PID控制算法,PID控制器,参数模型,神经网络,BP算法
Abstract
Classical PID control algorithm,as a general method of industrial process control,application scope is broad- principle, it does not depend on the specific mathematical model of the controlled plant,but tuning algorithm parameters is a very difficult more important,even if tuning the parameter pleted,as parameters do not have adaptive capacity,due to a change in environment,PID control of the response of the system deviation get worse,parameters need to be re- response to these problems,people have been using the adaptive method of fuzzy,works to adjust PID parameters,try hard to e this normal circumstances,an adaptive control system can be capable of running,and the corresponding parameters should adapt to tlle change in status of the scene,so the corresponding parameters must be based on the data of the scene to conduct online identification or -time—varying parameters can be confirmed for a period of on-line identification,but the time-varying parameters system will be necessary to continue this ongoing process,so the requirement of fast identification or the relative slow pace of change of parameters,greatly limits the application of adaptive e this limitation,this paper uses the ideology of literature[1],the technology of work will be used in the process of paramet