Year: 2026
Type: article
Abstract: Magnetic levitation systems (MLSs) exhibit complex nonlinear electromagnetic dynamics and significant variations across different operating points, which pose challenges for model-dependent controllers like PID and classical sliding mode control (SMC) in achieving high-precision performance. To tackle the inherent model uncertainty in MLSs, this study introduces an uncertainty-aware control framework that improves the robustness and efficiency of SMC using a radial basis function neural network (RBF NN). The RBF NN is capable of learning and adapting to nonlinearities, variations dependent on operating points, and lumped disturbances in real time. This adaptation alleviates the burden on the SMC’s robust term and reduces chattering. Experimental results from a laboratory MLS show that the RBF-enhanced SMC decreases the maximum closed-loop sensitivity by 21.1% and reduces sinusoidal tracking error by 80.2% compared to a conventional PID controller. These findings demonstrate that the uncertainty-aware adaptive control significantly enhances robustness, accuracy, and stability in magnetic levitation systems.
Authors: Seseung Namgung, Hyeong-Joon Ahn
Institution: Soongsil University
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: bronze
Awards: G1716110454, G5272650963
OpenAlex ID: W7140467410
FWCI:
Cites:
Cited by:
Related to: