A model-free dynamic sliding mode control approach with iterated-integral terms on sliding surface for chemical processes

This paper introduces a novel control approach called Model-Free Dynamic Sliding Mode Control with an Iterated-Integral Sliding Surface (MF-DSMC), designed to improve the control performance of chemical processes characterized by nonlinear dynamics, time-varying behavior, and system uncertainties. The proposed method integrates the robustness of Sliding Mode Control (SMC) with the adaptability of Model-Free Control (MFC), using an iterated-integral sliding surface to enhance transient response. A data-driven ultralocal model is employed in real time to estimate the system's behavior and adapt the switching gain dynamically. This reduces the dependence on fixed parameters and helps to mitigate chattering effects. The MF-DSMC strategy is validated through two thermal process case studies: (i) a simulated mixing tank with long and varying time delay, and (ii) an experimental setup using the Temperature Control Laboratory (TCLab) platform. The simulation assesses the controller's robustness and reference tracking under nonlinear and delayed conditions. The experimental setup evaluates its real-time performance. The results show that MF-DSMC achieves improved precision, adaptability, and robustness compared to conventional control methods. In particular, during the TCLab experiments, MF-DSMC obtained the better performance indices in terms of Integral of Absolute Error (IAE), Integral of Squared Error (ISE), and Total Variation of the control signal (TVu), outperforming a classical Model-Free Proportional-Integral (MF-PI) controller.