Design of Neuro-Controller and its Application to Process Control

神经控制器的设计及其在过程控制中的应用

• 批准号: 05650405
• 负责人: OMATU Sigeru
• 金额: $ 1.41万
• 依托单位: University of Tokushima
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for General Scientific Research (C)
• 财政年份: 1993
• 资助国家: 日本
• 起止时间: 1993 至 1994
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-05650405/
• 关键词: neuro-contrrol; adaptive control; Process control; learning control; PID control; temperature control; chemical process; generalization; ニューラルネットワーク; 階層型ニューラルネットワーク; 情報量; 学習曲線

Aim of the study is to design the neuro-controller and apply it to process control problems. In this study, we have designed three kinds of neuro-controllers, which are called series type, parallel type, and self-tuning type neuro-controllers. For the series type one, we have proposed the direct inverse control algorithm by using neural networks. In the parallel type one, we have introduced a feed-back error learning algorithm. As the last type, we have designed a self-tuning PID type neuro-controller where the proportional, intergral, and derivative gains are tuned in such a way that the error between plant output and desired output is minimized. After performing the neuro-controller designed here by simulations, we have determined the above PID gains by changing the structure of neural networks, Then we have applied the neuro-controller to real process control systems. First, we have applied the temperature control for water-bath and heating furnace and then applied the speed control of electric vehicle systems. From these applications, we have shown that proposed neuro-controller has worked well in real application problems.

本研究的目的是设计神经控制器，并将其应用于过程控制问题。在本研究中，我们设计了三种类型的神经控制器，分别为串联型、并联型及自校正型神经控制器。对于串联型系统，提出了利用神经网络的直接逆控制算法。在并行型学习算法中，我们引入了反馈误差学习算法。作为最后一种类型，我们设计了一个自校正PID型神经控制器的比例，积分和微分增益调整的方式，植物输出和期望的输出之间的误差最小化。在对所设计的神经网络控制器进行仿真后，通过改变神经网络的结构，确定了上述PID增益，并将神经网络控制器应用于真实的过程控制系统中。首先，我们应用了水浴和加热炉的温度控制，然后应用了电动车系统的速度控制。从这些应用中，我们已经表明，建议的神经控制器在真实的应用问题中工作得很好。

项目成果

タノマル ジュリオ,大松 繁: "Efficient On-Line Training of Multilayer Neural Controllers" 計測自動制御学会論文誌. 27. 272-280 (1993)

Julio Tanomaru、Shigeru Omatsu：“多层神经控制器的高效在线训练”仪器与控制工程师学会会刊 27. 272-280 (1993)。

Y.Rubiya,S.Omatu: "A Multivariable Self-Tuning PID Controller" International Journal of Control. 57. 1387-1403 (1993)

Y.Rubiya，S.Omatu：“多变量自整定 PID 控制器”国际控制杂志。

Marzuki Khalid,Sigeru Omatu,& Rubiyah Yusof: "MIMO Furnace Control with Neural Networks" IEEE Transactions on Control Systems Technology. 1(No.4). 238-245 (1993)

马祖基·哈立德、西格鲁·奥马图、

Marzuki Khalid,Sigeru Omatu,& Rubiyah Yusof: "Neural Networks for Self-Tuning Process Control Systems" Journal of Artificial Neural Networks. 1(No.1). 23-49 (1994)

马祖基·哈立德、西格鲁·奥马图、

松村茂憲,大松繁,樋笠博正: "PID型ニューラルコントローラによる電気自動車のトルク制御" 電気学会論文誌C分冊. 113(10号). 841-848 (1993)

Shigenori Matsumura、Shigeru Omatsu、Hiromasa Higasa：“使用 PID 型神经控制器的电动汽车的扭矩控制”，日本电气工程师协会学报，卷 C.113（第 10 期）（1993 年）。