Adaptive super twisting algorithm using gated recurrent unit blended RBF NN structure
Authors :- V. Mehra, D. Shah and A. Mehta
Publication :- Journal of Control and Decision (T&F), 2025
This paper proposes an adaptive Super Twisting Algorithm-based Second Order Sliding Mode Controller design that uses a novel Gated Recurrent Unit blended Radial Basis Function Neural Network (GRURBF NN) structure. The proposed structure has a GRU in the first hidden layer and an RBF in the second hidden layer. The GRU and RBF work together to improve the estimation of the unknown function šā”(ā ) in the nonlinear dynamical system. The estimated function is then used to design an adaptive Super Twisting controller and adaptive laws for updating Neural Network parameters. The closed-loop asymptotic stability and finite-time convergence of the system are also ensured using the Lyapunov theory. Finally, the controller is implemented on a two-link robotic arm with serial flexible joints to show the performance and efficacy. The experimental results exhibit that the proposed structure outperforms the traditional structure in tracking, robustness, efficient function approximation, and generation of smooth control signals.