BP算法在线优化PI控制器参数的PLC实现

文档编号:ZD155   附开题报告,任务书,文档,外文翻译,和其他过程材料  文档字数:12486,页数:31

摘  要
目前工业自动化水平已成为能够衡量各行各业现代化水平的一个重要标志。而常规PID是现在的工业控制中被广泛使用的一种控制方法。常规PID控制器具有结构简单、可靠性高的特点。本设计通过构建新的人工神经网络自学习PI控制器模型，将BP神经网络和常规PI 控制器融为一体，该模型既具有神经网络自学习、自适应及逼近任意函数的能力，又具有常规PI控制器结构简单、可靠性高的特点。本设计首先建立了单容水箱过程控制系统的对象，其次将BP神经网络在线优化PI控制器算法用PLC程序实现，并应用到单容水箱过程控制系统中，实验结果表明BP神经网络在线优化PI的实时监控曲线优于常规PI控制器的控制效果，系统控制性能指标中的调节时间和上升时间得到了较好的改善，证明了BP神经网络算法优化PI控制器的可行性和有效性。

关键词： PI控制器；神经网络BP算法；PLC
The PLC Realize of Optimal PI Controller Parameters on Line Base on BP Net
Abstract
At present the level of industrial automation has already become an important symbol, which is able to judge all trades modernized level. And now the conventional PID is a control method which is used widely in the industrial control. The routine PID controller has characteristic of the structure simplicity and the high of reliability. Combining the routine PI controller with BP neural networks, a new model of the PI controller that is able to self -learning of the new manpower neural networks is designed in this paper. Then the new model not only has the ability of self-learning neural network, adaptive arbitrary function approximation, but also has a characteristic of structure simple and high reliability in the conventional PID controller. Firstly this design built a object of single-tank process control system. Secondly the algorithm of PI controller parameters by optimization of BP neural networks on line is realized by PLC program in single-tank-testing process control devices. The results show that the algorithm of PI controller parameters by optimization of BP neural networks on line is better than the conventional PI controller. The adjusted time and the rise time is improved, and the algorithm is feasibility and effective.

Keywords：PID controller; BP Neural network algorithm; PLC

目  录
摘要 I
Abstract II
1 绪论 1
1.1 概述 1
1.2 设计方法及内容 3
2 基本理论 4
2.1 PID控制 4
2.2 BP人工神经网络学习算法 7
2.3 PLC工作原理 11
2.4 S7-200软件 13
3 程序设计 14
3.1 实验装置 14
3.2 I/O地址分配 14
3.3 PID回路表 15
3.4 变量地址统计 15
3.5 编程思想 16
3.6 主要程序介绍 18
3.7 实验数据及结果 24
结论 25
致谢 26
参考文献 27

相关论文