<p>As a typical mechatronic system, the dual-screw feed system (DSFS) requires fast response and high-precision position synchronization control, which remains a research hotspot. However, variations in the worktable’s center of mass, time-varying system states, and unknown disturbances present significant challenges for controller design. This paper aims to stabilize both the worktable position error and the beam rotation angle, proposing a fixed-time dynamic surface control method based on an immersion and invariance adaptive radial basis function neural network observer. First, a more precise rotational coupling dynamic model is established, considering the friction within the DSFS and the rotational dynamics caused by synchronization errors. Second, to better handle the unmodeled dynamics and unknown disturbances of DSFS, an immersion and invariance adaptive law is applied to improve the adaptive rate of the radial basis function neural network observer, breaking through the traditional equivalent weight update principle and further enhancing the estimation accuracy of uncertainties and external disturbances. Then, a variable-gain fixed-time dynamic surface control method is proposed. By constructing a variable-gain function that is inversely proportional to the error variation rate, the gain of the traditional fixed-time dynamic surface controller is dynamically adjusted, achieving fast convergence in large-error stages while ensuring high accuracy in small-error stages. Finally, the obtained controller output is allocated through coupling parameters to maintain a small rotation angle of the beam during motion while suppressing excessive coupled internal forces caused by beam rotation. Simulation and experimental results demonstrate the effectiveness of the proposed control method.</p>