Robust Magnetic Suspension Control via Nonlinear Disturbance Observer-Based Adaptive Fuzzy Neural Techniques

This paper presents a robust cascaded control strategy for an active magnetic bearing system based on a hybrid fuzzy–neural network controller integrated with a nonlinear disturbance observer. The proposed approach combines sliding mode control (SMC) with a proportional–integral–derivative (PID)-based sliding surface to enhance system stability and dynamic performance. To address uncertainties and external disturbances that are difficult to model mathematically, a disturbance observer is designed to estimate and compensate for unknown inputs using measured input–output signals. The estimated disturbances are further processed through a fuzzy neural logic system to improve robustness and reduce chattering effects commonly associated with sliding mode control. The fuzzy–neural network is structured in multiple layers, where the sliding surface acts as the input, and successive layers perform context processing, delayed signal handling, and nonlinear mapping to generate a smooth control output. This hierarchical architecture ensures adaptive learning and effective disturbance rejection while forcing chattering to approach zero. The performance of the proposed control scheme is validated through MATLAB/Simulink simulations, demonstrating improved suspension stability, fast convergence, and strong robustness against external disturbances. The results confirm the effectiveness of combining fuzzy–neural intelligence with nonlinear disturbance observer-based cascaded control for active magnetic bearing systems