Abstract: Welding technology is a critical manufacturing process in modern equipment production. However, welding deformation not only compromises dimensional accuracy and load-bearing capacity but also reduces service life and increases production costs, it has long been a major focus of both academic research and industrial practice. This paper provides a systematic review of significant advances in the study of welding deformation, and structures around a framework of mechanism, prediction and control. First, the mechanisms of welding deformation and the influences of material properties, structural characteristics and process parameters are analyzed. Subsequently, typical prediction methods, including experimental measurement, theoretical analytical models, numerical finite element simulation and intelligent prediction based on machine learning, are reviewed, with comparisons of their applicability and limitations. Then, techniques for controlling and correcting welding deformation are summarized, covering both active control and passive control approaches. Comprehensive analysis indicates that although substantial progress has been made in understanding deformation mechanisms and developing numerical simulations, challenges remain in efficient prediction, precise control and intelligent application for complex structures. In the future, integration of multiscale simulations with data-driven methods will offer promising avenues for both research and engineering applications in welding deformation management.