Time-delay Estimation Based Model-free Control with Adaptive Iterative Learning Compensator for Parallel Back-support Exoskeleton

To perform the trajectory tracking control of a 4-SPS/SC parallel structure based back-support exoskeleton for the repetitive rehabilitation, a time-delay estimation based model-free control with adaptive Iterative learning compensator and initial state learning (AILTDE-MFC) is proposed in this paper. Firstly, to assist with many types of waist movements and not limit the natural motion, a 4-SPS/SC parallel structure based back-support exoskeleton is designed in SolidWorks and the virtual prototype is built in MATLAB/Simscape Multibody. Secondly, for the AILTDE-MFC design, an ultra-local model is utilized to reconstruct the system dynamics and the lumped unknown information in the model is obtained by TDE, which can realize model-free trajectory tracking control in time domain. The designed iterative adaptive law is to compensate the estimation error of TDE and the initial state learning is to guarantee the convergence of initial states. The stability of AILTDE-MFC is verified by CEF. Finally, to validate the proposed method, a visualized co-simulation is implemented by a virtual prototype. Compared with PD and TDE-iPD controllers, the proposed algorithm demonstrates the high tracking performance in time domain and the error convergence along iteration axis.