Data driven sliding mode control with minimum operator and its application on a magnetorheological fluid dual clutch

Rate-dependent hysteresis and strong non-linearity have posed significant challenges to the modeling and control of magnetorheological fluid devices, such as magnetorheological fluid clutches for electric vehicle transmission systems, where these clutches must operate in transient states for effective torque tracking. Therefore, in this paper, a modified partial-form dynamic linearization (mPFDL) data model is proposed to describe the relationship between input current and output torque of a magnetorheological fluid dual clutch (MFDC), where an adaptive mechanism is designed to achieve real-time estimation of the uncertain parameter in the mPFDL data model, while the disturbances and unmodeled dynamics are collectively treated as a total extended state that is estimated by a discrete-time extended state observer. Based on the mPFDL model, a minimum operator (MO) based discrete time sliding mode controller is designed to control the transmission torque of the MFDC. Furthermore, the proposed MO-based reaching law limits the rate of variation of the sliding mode variable when it is far from the sliding manifold, and the sliding mode variable remains within the quasi-sliding mode domain in a finite number of steps, as demonstrated by the theoretical analysis. Finally, numerical simulations are performed to verify the effectiveness of the proposed control scheme, and experiments based on the MFDC further validate the theoretical findings. The experimental results confirm that the proposed control method can achieve better torque tracking performance than the compared super-twisting algorithm and data-driven control method in both steady and transient states.