Adaptive neural network predefined time control for magnetic levitation linear motor with current constraints and mismatched disturbances

To address the issues of the overcurrent protection and mismatched disturbance in magnetic levitation linear motor (MLLM), an adaptive neural network predefined time control is proposed. Initially, the operational mechanism of the MLLM is analyzed and a mathematical model is established. Subsequently, a nonlinear mapping technique is employed to convert the constrained current variable into an unconstrained variable. An adaptive neural network is integrated into the backstepping method to estimate and compensate for the mismatched disturbance, and a differentiator is designed to handle the explosion of complexity problem. An adaptive law is formulated to mitigate the estimation errors of the neural network and the differentiator. Using Lyapunov stability theory and predefined time stability theory, it is demonstrated that the tracking error can converge to a neighborhood near zero in a predetermined time with a simple parameter tuning, and the current fails to exceed a predetermined range. Finally, experiments show that the proposed controller exhibits robustness and tracking performance while adhering to current constraints.