Abstract: To address the challenges in university practical education of engineering machinery—including insufficient large-scale equipment, difficulties in full-staff hands-on training, high safety risks, and the disconnect between skill development and physical/psychological well-being—this paper proposes a dual-dimensional collaborative training framework integrating “operational behavior and physiological state”. By incorporating physiological metrics into the evaluation system, it establishes an adaptive virtual simulation platform that combines virtual simulation experiments with multi-channel physiological data collection. This platform creates a closed-loop mechanism: “immersive operation → data collection → analysis and feedback → adaptive regulation → operational optimization”, which enables dynamic adjustment of virtual scenario difficulty and delivers personalized guidance. Taking the virtual operation of excavators as an example, personalized adaptive regulation is implemented based on the operator’s physiological data thresholds to ensure the coordinated improvement of students’ operational skills and physiological stability. This achieves a transition from an “experience-driven” to a “data-driven” teaching model, providing references for the reform of practical teaching in mechanical engineering disciplines as well as platform support for human-computer interaction experiments in fields such as ergonomics.