An Innovative MIMO Iterative Learning Control Approach for the Position Control of a Hydraulic Press

To improve the performance of hydraulic press position control and eliminate the need to manually define control signals, this paper proposes a multi-input-multi-output (MIMO) Iterative Learning Control (ILC) algorithm. The MIMO ILC algorithm design is based on the inversion of the known low frequency dynamics of the hydraulic press, whereas the unknown and uncertain high frequency dynamics are discarded due to their low influence in the learning transient. Moreover, for the MIMO ILC convergence condition, a graphical method is proposed, in which the ILC learning filter eigenvalues are analyzed. This method allows studying the stability and convergence rate of the algorithm intuitively. Theoretical analysis and results prove that with the MIMO ILC algorithm the position control is automated and that high precision in the position tracking is gained. A comparison with other model inverse ILC approaches is carried out and it is shown that the proposed MIMO ILC algorithm outperforms the existing algorithms, reducing the number of iterations required to converge while guaranteeing system stability. Furthermore, experimental results in a hydraulic test rig are presented and compared to those obtained with a conventional PI controller.