| 一种基于神经网络的学习控制方法 |
| |
| 引用本文： | 张冰 张基宏. 一种基于神经网络的学习控制方法[J]. 深圳大学学报(理工版), 1997, 14(1): 83-89 |
| |
| 作者姓名： | 张冰 张基宏 |
| |
| 作者单位： | 深圳大学电子工程系 |
| |
| 摘 要： | 提出一种新的基于神经网络的增强式学习控制方法．学习控制器包括系统性能的评估部分及由性能评估提供的增强信号引导下进行学习的神经网络部分．模拟及物理实验结果表明，此方法具有学习速度快，适应性强，通用性好等特点．
|
| 关 键 词： | 神经网络；BP算法；增强学习；智能控制；机器学习 |
A Neural Network Approach to Reinforcement Learning Control |
| |
| Zhang Bing and Zhang Jihong Dept.of Electronic Engineering Shenzhen University Shenzhen P.R China. A Neural Network Approach to Reinforcement Learning Control[J]. Journal of Shenzhen University(Science &engineering), 1997, 14(1): 83-89 |
| |
| Authors: | Zhang Bing and Zhang Jihong Dept.of Electronic Engineering Shenzhen University Shenzhen P.R China |
| |
| Affiliation: | Zhang Bing and Zhang Jihong Dept.of Electronic Engineering Shenzhen University Shenzhen 518060 P.R China |
| |
| Abstract: | A new scheme of reinforcement learning control using neural networks is presented.The learning controller consists of a perfonnarce evaluation part and a neural network part that learns under the influence of the reinforcement signal provided by performance evaluation.Results of simulation and physical experiments showed that the learning controller was able to learn to control a nonlinear system at a much faster learning rate and more adaptive to system parameter changes than most of the other learning control algorithms for the same task.Furthermore,it is general-purpose. |
| |
| Keywords: | neural net works BP method reinforcement learning |
| 本文献已被 CNKI 维普 等数据库收录！ |
