Dynamic hysteresis compensation and iterative learning control for underwater flexible structures actuated by macro fiber composites.

Studies on the hydrodynamics of underwater flexible structures are gaining significant attention. To improve the performance and stability of underwater flexible structures in complex water environments, new control strategies must be proposed. This study proposes underwater rate-dependent hysteresis compensation and iterative learning control (ILC) of the flexible structure actuated by macro fiber composites (MFC). The underwater rate-dependent hysteresis model of the MFC partially-actuated structure consists of two components: a modal state space model characterizing the underwater rate-dependent behavior, and a modified Prandtl–Ishlinskii (MPI) model describing the bias bipolar hysteresis. The MPI model is combined with a PI and a polynomial model of fractional orders. To obtain a unique analytic inverse model, constraints for the inverse are proposed. Results show the varying bias ellipse-like hysteresis behaviors are well captured. Tracking performances are greatly improved with the feedforward compensator. Furthermore, an ILC is developed to address model uncertainties and unsteady hydrodynamics. The convergence and robustness of the ILC are discussed. Parameters of the zero-phase and learning filters are determined. Experimental results demonstrate the tracking errors of the underwater structure are significantly reduced. These findings are meaningful for the realization of flexible structures for marine engineering, underwater robots and other fields. • Underwater rate-dependent hysteresis compensation and iterative learning control are proposed. • A modified Prandtl-Ishlinskii hysteresis model combines a PI model with polynomials of fractional orders. • Constraints for the unique analytic inverse of polynomials of fractional orders are defined. • A modal state space model is developed to characterize the underwater rate-dependent behavior. • A zero-phase low-pass filter is proposed to address the phase lag caused by the Q -filter. [ABSTRACT FROM AUTHOR]