Data-driven Iterative Learning Model Predictive Control for Pneumatic Muscle Actuators

Iterative learning control (ILC) has been considered as a promising alternative for the control of pneumatic muscle actuator (PMA). However, this controller suffers from a challenge that it is difficult to deal with the complex nonlinear characteristics of PMA. To solve this problem, a novel iterative learning model predictive control (ILMPC) approach, by utilizing the data-driven model, is designed and analyzed in this article. Firstly, the dynamics of PMA is converted into Takagi-Sugeno (T-S) fuzzy nonlinear auto-regression with exogenous inputs (NARX) model, and the differential evolution (DE) estimation algorithm is applied to estimate parameters of the NARX model by utilizing the input and output data. Secondly, the controller of ILMPC is designed and the convergence performance of the controller is verified through theoretical analysis. Finally, the capability of this control method is confirmed via experimental study. Experimental results demonstrate that the proposed ILMPC can achieve satisfactory tracking control and exhibits robustness against load varying.