AI赋能新工科跨学科项目式教学资源体系建设与实践研究

Research on the Construction and Practice of an AI-Empowered Interdisciplinary Project-Based Learning Resource System for Emerging Engineering Education

摘要: 新一轮科技革命和产业变革对工程人才培养提出了智能化、跨学科和实践化的新要求。新工科教育强调面向未来产业需求，推动人工智能技术、专业知识与真实工程问题深度融合。项目式学习作为培养学生复杂工程问题解决能力和创新实践能力的重要方式，与新工科建设目标高度契合。然而，当前跨学科项目式教学仍存在资源整合不足、实践场景有限、个性化指导薄弱和评价方式单一等问题。基于此，本文从政策逻辑、技术逻辑和知识逻辑三个维度分析AI赋能新工科跨学科PBL的内在机理，并提出由模块化课程群资源、“双映射”数字孪生实践平台和生成式AI学习支架构成的教学资源体系。研究认为，该体系有助于促进产教融合，提升学生工程实践能力、创新能力和跨学科素养，但在应用中仍需关注伦理安全、技术依赖、教师数智素养和数字鸿沟等问题。在此基础上，本文进一步提出由物理实体与资源层、数据接入与治理层、孪生建模与仿真层、决策回写与安全门控层、教学服务与评价层构成的技术架构，并以智慧交通项目说明仿真结果如何转化为可执行参数，反向指导实体设备调试与工程验证；同时将生成式AI学习支架设计为由检索增强生成知识库、学习者模型、项目阶段识别、多角色智能体和支架渐隐机制组成的可追踪系统。理论上，研究融合认知学徒制、支架理论、自我调节学习、认知负荷与分布式认知，明确“真实任务–双向表征–人机协同–反思调节–迁移创造”的作用链条。

Abstract: The new round of scientific and technological revolution and industrial transformation has raised new requirements for engineering talent cultivation, emphasizing intelligence, interdisciplinarity, and practice orientation. Emerging Engineering Education focuses on future industrial needs and promotes the deep integration of artificial intelligence technologies, professional knowledge, and real engineering problems. Project-Based Learning, as an effective approach to cultivating students’ ability to solve complex engineering problems and engage in innovative practice, is highly consistent with the goals of Emerging Engineering Education. However, current interdisciplinary PBL still faces problems such as insufficient resource integration, limited practical scenarios, weak personalized guidance, and a single evaluation approach. Therefore, this paper analyzes the internal mechanism of AI-empowered interdisciplinary PBL in Emerging Engineering Education from the perspectives of policy logic, technical logic, and knowledge logic. It further proposes a teaching resource system composed of modular curriculum resources, a dual-mapping digital twin practice platform, and a generative AI-driven learning scaffold. The study argues that this system can promote industry-education integration and improve students’ engineering practice ability, innovation capacity, and interdisciplinary competence. Meanwhile, issues such as ethical security, technological dependence, teachers’ digital and intelligent literacy, and the digital divide should be carefully addressed. On this basis, this paper further proposes a technical architecture consisting of five layers: The physical entity and resource layer, the data acquisition and governance layer, the digital twin modeling and simulation layer, the decision feedback and safety control layer, and the teaching service and evaluation layer. Taking a smart transportation project as an example, it illustrates how simulation results can be transformed into executable parameters, thereby providing reverse guidance for the debugging of physical equipment and engineering validation. Meanwhile, the generative AI learning scaffold is designed as a traceable system composed of a retrieval-augmented generation knowledge base, learner model, project-stage identification mechanism, multi-role intelligent agents, and scaffold fading mechanism. Theoretically, this study integrates cognitive apprenticeship theory, scaffolding theory, self-regulated learning theory, cognitive load theory, and distributed cognition theory, establishing an operational pathway of “authentic tasks-bidirectional representation-human-AI collaboration-reflective regulation-transfer and creation.”