A Predefined‐Time Zeroing Neural Network Model for Solving Multi‐Linear Systems

ABSTRACT This paper presents a zeroing neural network model incorporating a novel activation function for solving multi‐linear systems. Its key feature is a tunable parameter that allows users to directly preset the convergence time, offering greater flexibility than fixed‐time approaches. Theoretical analyses and comparisons confirm that the proposed model achieves predefined‐time convergence and exhibits stronger robustness to noise disturbances than existing models. Numerical simulations confirm that it outperforms existing models in both convergence speed and computational efficiency. This method shows practical potential for real‐time applications such as robotic control and dynamic parameter estimation.