An AI‐Driven Framework for Evaluating the Quality of Mechanical Engineering Undergraduate Capstone Topics: Development and Empirical Validation

Undergraduate capstone projects were a critical component of mechanical engineering talent cultivation, as the quality of topic selection directly determined both the overall quality of the project and the effectiveness of students' competency development. However, the absence of scientific and operable evaluation criteria for topic selection review led to formalistic assessments, ultimately compromising the quality of graduation projects. To address this issue, this study established an evaluation system for mechanical engineering undergraduate capstone project topics based on a dual‐dimensional subjective–objective assessment model. A total of 150 graduation projects from the 2024 cohort of the Mechanical Engineering program at Wuhan University of Technology were used as samples. The effectiveness of the proposed evaluation system was validated using the DeepSeek AI model. The results indicated that the predicted topic quality grades generated by the evaluation system achieved a 73.3% consistency rate with students' actual graduation project performance grades. These findings demonstrated that the proposed system could effectively enhance the standardization and reliability of topic selection reviews, providing a practical and valuable theoretical framework and tool for ensuring the quality of undergraduate capstone projects.