International Journal of Advances in Electrical Engineering
  • Printed Journal
  • Refereed Journal
  • Peer Reviewed Journal

P-ISSN: 2708-4574, E-ISSN: 2708-4582

International Journal of Advances in Electrical Engineering


2025, Vol. 6, Issue 1, Part A
Neural network-driven dynamic surface control for nonlinear systems: Integration of zero-tracking error mechanisms and experimental validation


Author(s): Nino Abashidze, Levan Tsiklauri, Tamar Chikhladze, Giorgi Davitadze and Mariam Abuladze

Abstract: Neural network-based control strategies have emerged as a powerful approach for handling complex nonlinear dynamic systems across various engineering domains, including robotics, aerospace, and industrial automation. Traditional control methodologies, such as PID and adaptive controllers, often fail to provide high accuracy under uncertain and dynamically changing conditions. To address these limitations, this study proposes a Neural Network-Driven Dynamic Surface Control (DSC) framework that integrates zero-tracking error mechanisms to improve tracking precision and robustness. The primary objective is to design a control system that ensures high stability, adaptability, and computational efficiency while maintaining real-time control accuracy.The proposed framework was evaluated through an experimental setup involving a DC servo motor system, a six-degree-of-freedom robotic arm, and a nonlinear pendulum system, interfaced with a Texas Instruments DSP processor and an NVIDIA GPU module for neural network computations. The control algorithm was designed with a feedforward neural network (FNN) predictor and a Lyapunov-based adaptive controller, trained using an online gradient descent approach. The system was tested under sinusoidal disturbances, Gaussian noise, and step input perturbations, and tracking performance was measured using Root Mean Square Error (RMSE) and Integral of Time-Weighted Absolute Error (ITAE) metrics.The results demonstrated that the proposed DSC framework achieved superior tracking accuracy, with RMSE as low as 0.012 under baseline conditions and minimal error deviations under varying disturbances and load conditions. Statistical validation through ANOVA tests (p > 0.05) confirmed the consistency of system performance. The study concludes that integrating neural network-based learning with DSC significantly enhances control accuracy, real-time adaptability, and robustness, making it suitable for autonomous robotics, aerospace, and industrial automation. Further research should focus on optimizing computational efficiency, hybrid control strategies, and real-world hardware implementations.

DOI: 10.22271/27084574.2025.v6.i1a.82

Pages: 31-35 | Views: 84 | Downloads: 36

Download Full Article: Click Here

International Journal of Advances in Electrical Engineering
How to cite this article:
Nino Abashidze, Levan Tsiklauri, Tamar Chikhladze, Giorgi Davitadze, Mariam Abuladze. Neural network-driven dynamic surface control for nonlinear systems: Integration of zero-tracking error mechanisms and experimental validation. Int J Adv Electr Eng 2025;6(1):31-35. DOI: 10.22271/27084574.2025.v6.i1a.82
International Journal of Advances in Electrical Engineering
Call for book chapter