Modeling and vibration control with neural network for flexible multi-link structures

Vibration control for flexible structures has been studied by many researchers. We also have proposed decentralized vibration control for flexible structures by decoupling the mode quantities of other links. However if the modeling errors exist in the control system, the performance of the decoupling control goes down since the control method is based on the mathematical model. In this paper, in order to consider the modeling errors, adaptive control based on a neural network is applied to the vibration control of a flexible multi-link system. And by considering the axial forces in modeling, it is found that the modal dynamics is given as a nonlinear system. The effectiveness of the proposed control method is shown by the simulations of a 5-link robotic arm.