Learner Behavior Recognition and Dynamic Guidance Mechanism in Interactive Mobile Simulation Systems for Vocational Education

Mobile simulation technologies provide scenario-based solutions for practical instruction in vocational education. However, existing systems are commonly constrained by limited interaction perception modalities, delayed learner behavior recognition, and a lack of personalized guidance strategies, resulting in insufficient adaptability to the acquisition of complex vocational skills. To address these challenges, a learner behavior recognition and dynamic guidance mechanism integrating multimodal perception and imitation learning was proposed, and a three-layer edge-cloud collaborative architecture consisting of perception, understanding, and guidance was established. Multimodal interaction perception was achieved through the synchronous acquisition of user interface (UI) layout trees, device sensor data, and operation sequence data. A lightweight spatiotemporal attention network was designed to perform behavior encoding and recognition, while imitation learning was introduced to enhance recognition performance in long-tail operation scenarios. A context-aware guidance decision engine was constructed using reinforcement learning, enabling the dynamic generation of guidance strategies adapted to learners’ skill levels and task contexts. Deployment efficiency on mobile devices was improved through dynamic computation offloading. Experimental results demonstrated that the proposed lightweight spatiotemporal attention model achieved a recognition accuracy of 94.2% for basic operations and a macro-F1 score of 0.876 for long-tail operations, with end-to-end latency consistently maintained at or below 325 ms. The dynamic guidance mechanism reduced average task completion time by 28.3% and decreased operation error rates by 41.2%. In addition, overall system deployment performance significantly outperformed that of existing mobile educational systems. These findings validate the effectiveness and superiority of the proposed approach and provide a practical and efficient technical framework for the deep integration of mobile computing technologies with vocational education, substantially enhancing vocational skill acquisition efficiency.