摘要
针对轧机液压弯辊系统存在非线性、时变性等特点,采用基于前馈神经网络的内模控制方法,将优化网络用于神经网络辨识器和内模控制器的离线训练,采用学习率自适应调整的改进BP算法在线调整网络权值。仿真研究表明,将优化网络用于液压弯辊系统控制,提高了液压弯辊系统的动态响应速度和稳态跟踪精度,具有较强的快速性和鲁棒性,能够取得理想的控制效果。
The internal model control using neural network was introduced in allusion to mill hydraulic bending roll system's nonlinear,time-varying performance. Author proposed means for using optimal neural network to train identificator and internal model controller off-line, improved adaptive BP algorithms' learning law on-line. The simulation results have demonstrated that this kind of controller improves dynamic response speed and tracing accuracy. The hydraulic bending roll system possesses rapid response and strong robustness, the control effect is ideal.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2007年第20期2419-2421,共3页
China Mechanical Engineering
基金
燕山大学博士基金资助项目(B70)
关键词
内模控制
神经网络
模型辨识
BP算法
液压伺服系统
internal model control
artifical neural network
model identification
BP algorithm
hy-draulic servo system