Neural network controller design for position control system improvement
This project focused on development of precise position control with a DC motor as an actuator using neural network controller. Neural network controller develop is proposed to overcome the problem of conventional controller weaknesses. Neural network controller is implemented using backpropagati...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Thesis |
| 语言: | English English English |
| 出版: |
2013
|
| 主题: | |
| 在线阅读: | http://eprints.uthm.edu.my/6671/1/24p%20MOHAMAD%20SYAH%20RIZAL%20ABDULLAH.pdf http://eprints.uthm.edu.my/6671/2/MOHAMAD%20SYAH%20RIZAL%20ABDULLAH%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/6671/3/MOHAMAD%20SYAH%20RIZAL%20ABDULLAH%20WATERMARK.pdf |
| 标签: |
添加标签
没有标签, 成为第一个标记此记录!
|
| 总结: | This project focused on development of precise position control with a DC motor as
an actuator using neural network controller. Neural network controller develop is
proposed to overcome the problem of conventional controller weaknesses. Neural
network controller is implemented using backpropagation training algorithm. Neural
network has ability to map unknown relationship input/output system and also nonlinear
system. To have knowledge about the system, the neural network is trained
using existing controller on the position control system, in this case PID controller.
On the training process, neural network controller and PID controller are having
same inputs, which are errors. After that, the outputs are compared and the delta of
them will used to adjust the network weight until the delta value in the acceptance
level. Then, neural network controller is set convergence. At this time, neural
network controller ready use to replace PID controller to control the system. To
interface between computer where neural network controller is embedded with the
DC motor as a position controller system actuator are done using RAPCON platform.
Based on the experimental results, show that neural network controller has better
performance with the rise time ( ) is 0.02s, the peak time ( ) is 0.05s, settling time
( ) is 0.05s, and percentage overshoot ( ) is 2.0%. |
|---|