Driven by the dual forces of “AI + Education” and the development of Emerging Engineering Education, electrical engineering courses face challenges characterized by intensive interdisciplinary knowledge, strong engineering orientation, and heightened demands for personalization. Electrical Machinery, as a core foundational course for electrical engineering majors, features a complex knowledge system and requires tight integration of theory and practice. Traditional teaching methods struggle to meet the needs for structured knowledge presentation, resource integration, and personalized guidance. Based on its specific characteristics, this study constructs a multimodal knowledge graph to achieve networked associations between disciplinary knowledge and engineering resources. By employing machine learning to collect and analyze student data related to theory, practice, and engineering tasks, a multi-dimensional learning evaluation model is established and integrated throughout the blended learning process. Practice has shown that this model strengthens students’ systematic understanding of knowledge, enhances their ability to apply theory and engage in engineering practice, assists teachers in precise instruction and dynamic resource iteration, provides an operable pathway for the deep integration of AI and electrical engineering courses, and holds promotional value for the intelligent reform of engineering-oriented courses.