Data-Driven Sliding Mode Control for Partially Unknown Nonlinear Systems

Abstract

This paper presents a novel data-driven control framework for multi-input multi-output (MIMO) nonlinear systems with partially unknown dynamics and bounded disturbances. In practical scenarios, exact cancellation of system nonlinearities is often infeasible due to limited model information and unpredictable disturbances. To address this challenge, sliding mode control (SMC) is adapted to a data-driven setting to ensure robust stability. The proposed framework comprises two key components: a data-driven robust controller that drives the system state trajectories to a predefined sliding surface, and a nominal controller, synthesized via data-driven semidefinite programming (SDP), to maintain stable sliding motion. Extensive simulation studies demonstrate that the proposed method outperforms state-of-the-art data-driven control approaches based on approximate nonlinearity cancellation, as well as the classical Deep Q-Network (DQN) reinforcement learning algorithm, achieving superior robustness and stabilization performance.