Abstract
This research work is based on a digital twin model, emphasizing the integration of digital twin models within a comprehensive five-dimensional framework. It introduces a breakthrough evaluation standard for digital twin models, addressing challenges related to construction, performance evaluation, adaptability, and value. The study specifically focuses on the use of Convolutional Neural Networks (CNN) for gesture recognition in Virtual Reality (VR) systems, emphasizing the importance of reliability analysis in aviation maintenance scenarios. This research presents a VR multi-source classification model developed with Python and PyTorch, which demonstrates exceptional stability and accuracy, exceeding 98%, across different environments. Key metrics such as precision, recall, and specificity are used to underscore the transformative impact of digital twin technology on the optimization of real-world entities, particularly in the field of aviation maintenance. The paper concludes by validating the model’s performance with experimental results.