|Table of Contents|

A PID-like Neural Network Control Based on Multi-step Prediction(PDF)

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

Issue:
2000年03期
Page:
199-202
Research Field:
Publishing date:

Info

Title:
A PID-like Neural Network Control Based on Multi-step Prediction
Author(s):
NiuYugang YangChengwu ZhaoJiancong ①
School of Power Engineering,NUST,Nanjing 210094)
Keywords:
neural netw ork predict ion robustness PID regulator
PACS:
TP18
DOI:
-
Abstract:
A mult-i step predict ion is presented in this paper based on PID-like neural netw ork cont rol scheme similar to the position successive PID in cont rolling mechanism. T he new cont roller can not only carry out the operat ions of proport ion, integral and different ial of error, but also adapt ively adjust PID parameters. The cont rol signal produced by this controller is a nonlinear combinat ion of the values of proportion, integral and derivat ive of errors. Hence, the cont radict ion of fast response and overshoot , which exists in the conventional PID regulator, can be effect ively solved. The new PID-like neural network cont roller is t rained by a performance index based on mult-i step predictive errors. This can effect ively overcome the shortcoming that cont rol signal severely fluctuates accompanied w ith one-step predict ion. Simulation result s prove that this new mult-i step prediction based on PID-like neural netw ork control system can effectively at tenuate random noise interference and is more robust and adaptive.

References:

1 王耀南, 童调生, 蔡自兴1 基于神经网络的智能PID控制及应用.信息与控制, 1994, 23( 3) :185~ 189
2 谭永红.神经网络自适应PID 控制及其应用.模式识别与人工智能, 1993, 6( 1) : 81~ 85
3 靳其兵, 顾树生. 一种参数快速收敛的多变量系统神经网络PID 控制.控制与决策, 1998,13( 7) : 448~ 452
4 李奇, 李世华. 一类神经网络智能PID 控制算法的分析与改进. 控制与决策, 1998, 13(4) : 311~ 316
5 韩京清.非线性PID 控制器. 自动化学报, 1994, 20( 6) : 487~ 490

Memo

Memo:
-
Last Update: 2013-03-25