Output tracking of flexible mechanisms by learning

Output tracking of flexible mechanisms by means of learning algorithms is investigated. The spillover effect which may adversely affect the control of flexible structures, is examined. It is shown that high frequency dynamics can be neglected with minor detriment to system performance. The learning algorithm introduced is of the Newton-Raphson type and its convergence is proved for the case of slow rigid motion. The performance of the proposed learning algorithm and the possibility of improving algorithm robustness by state feedback are illustrated by means of an example