神经网络PID控制算法的MATLAB仿真及其在环境控制中的应用学士学位论文.doc

本科毕业设计（论文） 课题名称 神经网络PID控制算法的MATLAB仿真及其在环境控制中的应用 摘要 PID（比例-积分-微分）控制器作为最早实用化的控制器已有50多年历史，现在仍然是应用最广泛的工业控制器。PID控制器简单易懂，使用中不需精确的系统模型等先决条件，因而成为应用最为广泛的控制器。PID控制室最早发展起来的控制策略之一，由于其算法简单、鲁棒性好和可靠性高，被广泛应用于过程控制和运动控制中，尤其适用于可建立精确数学模型的确定性控制系统。但是常规的PID控制器参数往往整定不良、性能欠佳，对运行工况的适应性很差。而神经网络具有很强的非线性映射能力、自学习能力、联想记忆能力、并行信息处理方式及优良的容错性能。 本课题设计提出就是为了建立一种单神经网络的PID控制器，使得人工神经网络与传统PID控制相结合互相补充，共同提高控制质量，并利用Matlab软件进行仿真。 关键词：控制理论；神经网络；pid控制；BP算法；神经网络pid；MATLAB仿真 ABSTRACT The PID ( PID ) controller as the first practical controller has 50 years of history, is still the most widely used industrial controller. The PID controller is simple and easy to understand, without the use of accurate system models prerequisites, and thus become the most widely used controller. The PID control room is the earliest developed one of the control strategy, because of its simple algorithm, good robustness and high reliability, is widely used in process control and movement control, especially can be applied to establish the precise mathematical model of uncertainty control system. But the conventional PID controller parameters often setting bad, poor performance, the operating condition adaptability is poor. While the neural network has very strong nonlinear mapping ability, self-learning ability, the capacity of associative memory, parallel information processing and fine fault-tolerant performance. This topic design proposed is to establish a single PID neural network controller, the artificial neural network and traditional PID control are combined to complement each other, work together to improve the control quality, and the use of Matlab software simulation. Key words: control theory; neural network; PID control; BP algorithm; neural network PID; MATLAB simulation 目录 第一章 绪论 - 5 - 1.1课题研究背景 - 5 - 1.2课题研究意义 - 6 - 1.3课题目前研究现状 - 8 - 1.4本文的主要任务及研究内容 - 10 - 第二章 神经网络 - 12 - 2.1神经网络的基本概念 - 12 - 2.2人工神经元模型 - 14 - 2.3神经网络的结构 - 16 - 2.4神经网络的工作方式 - 18 - 2.5神经网络的学习 - 18 - 2.6小结 - 20 - 第三章 PID控制器