Hysteresis modeling of magnetorheological fluid-based (MR) brakes with deep learning

Magnetorheological fluid-based (MR) brakes have been preferred in the design of an actuation system for a kinesthetic type of haptic device due to their high continuous resistive force/torque capacities per their size. Modeling the hysteresis input–output relationship of MR brakes is still an important challenge. We propose a deep learning-based approach by which the forward hysteresis behavior of an MR brake can be modeled. Two deep learning (DL) methods, long short-term memory (LSTM) and gated recurrent unit (GRU), which enable sequence-sequence modeling, are employed in this work. Moreover, a data pre-processing step is proposed to increase the variety of input signals. The performances of the aforementioned DL methods are evaluated by calculating their RMSE and coefficients of determination ( R 2 ) values on a different set of test signals.