Accurate Position Control of a Servo-Hydraulic Test Cylinder by Iterative Learning Control Technique

This paper describes a solution for repetitive position control problems with servo-hydraulic test cylinders. By extending the classical P-type iterative learning control algorithm with a frequency domain filtering technique, we obtain robust control performance with fast system convergence. In traditional P-type iterative learning control, there is always a contradiction between fast convergence and algorithm stability. With a large learning parameter value the control system is quickly convergent nevertheless in iteration domain the algorithm is unstable even if a stable feedback control system is applied; while with a small learning parameter value the algorithm is stable however the convergence rate is not satisfying. Extending ILC with a frequency domain low-pass filter allows us for achieving quick convergence rates during consecutive control iterations while still ensuring overall system stability. Based on a full simulation of the servo-hydraulic cylinder system we are able to simulate the ILC cylinder system behavior for our test setup, enabling for system performance investigation and optimization prior to the realization of a certain test setup. Experimental results with a real-world single-cylinder test rig validate our accurate system simulation environment and furthermore demonstrate the high position control performance of the proposed ILC method.