| 基于BP神经网络的PID控制器研究 |
| |
| 引用本文： | 李广军,张晶,曾安平. 基于BP神经网络的PID控制器研究[J]. 计算机仿真, 2009, 26(9): 128-131 |
| |
| 作者姓名： | 李广军 张晶 曾安平 |
| |
| 作者单位： | 宜宾学院网管中心,四川,宜宾,644007;宜宾学院网管中心,四川,宜宾,644007;宜宾学院网管中心,四川,宜宾,644007 |
| |
| 摘 要： | BP神经网络的PID控制器依赖于灵敏度信息实现参数在线调整,获得灵敏度信息非常重要。利用符号函数来获得灵敏度信息,计算不精确,利用RBF对被控对象在线辨识获得灵敏度信息,收敛速度慢。提出了一种新方法,即在RBF神经网络辨识的基础上,当误差较大时,利用符号函数获得灵敏度信息,以加速收敛;当误差较小时,利用RBF神经网络在线辨识获得灵敏度信息,以提高控制精度。仿真结果表明算法收敛速度快、精度高,控制效果优于符号函数实现的参数调整方法。
|
| 关 键 词： | 反向传输神经网络 符号函数 径向基神经网络 灵敏度信息 |
PID Control Based on BP Neural Network |
| |
| LI Guang-jun,ZHANG Jing,ZENG An-ping. PID Control Based on BP Neural Network[J]. Computer Simulation, 2009, 26(9): 128-131 |
| |
| Authors: | LI Guang-jun ZHANG Jing ZENG An-ping |
| |
| Affiliation: | Network Management Center;Yibin University;Yibin Sichuan 644007;China |
| |
| Abstract: | A BP neural network is utilized to achieve PID parameter self-adjustment based on Jacobian values.,which depends on a RBF neural network identification with slow convergence or on sign function with low precision.To gain Jacobian values,an algorithm is proposed in which a sign function is used based on a RBF neural network identification on line as bigger error and a RBF neural network identification is used as minor error. Results indicate that the effectiveness of this controller with high precision and f... |
| |
| Keywords: | |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
