Neural Network Control of Underwater Vehicles Based On Robust Learning Algorithm
Liang Peng etc. successfully applied neural network technologies to motion control of underwater robots, but low response speed and sensitivity to external noises limit the development in practice. In order to improve the performance of neural network controller, a robust learning algorithm is proposed. Firstly, the stable robust learning algorithm is presented, which is based on variable structure control theory. Secondly, the global stability conditions are explained in detail. Finally, in order to evaluate the performance, the control system based on the proposed algorithms is applied to the simulation platform of General Detection Remotely Operated Vehicle (GDROV).The results show that it not only possesses the good robustness to changing of learning-ratio and external noises but also can keep learning of neural network fast and stable, which is feasible to be applied to real time control of underwater vehicle.
Neural Network Robust Learning Algorithm Variable Structure Learning-ratio General Detection Remotely Operated Vehicle (GDROV)
Yushan Sun Xiao Liang Lei Wan Shaoji Fang
College of Shipbuilding Engineering, Harbin Engineering University Harbin, Heilongjiang Province 150001/China
国际会议
杭州
英文
815-819
2006-10-12（万方平台首次上网日期，不代表论文的发表时间）