“问题链 + AI”驱动的《冶金传输原理》教学模式重构——以高炉炉缸侵蚀为例

Reconstruction of “Metallurgical Transport Principles” Teaching Model Driven by “Problem Chain + AI”—Taking Blast Furnace Hearth Erosion as an Example

摘要: 传统工科教学常陷入理论抽象、工程脱节、反馈滞后的三重困境，在《冶金传输原理》等专业核心课程中尤为突出。虽然课程内容覆盖傅里叶定律等经典理论，却往往与“高炉炉缸侵蚀”等真实工程问题相割裂，学生难以实现从理论到工程应用的认知跃迁。为此，本研究以认知学徒制与情境学习理论为依据，提出“问题链 + AI工具”协同教学模式：以高炉炉缸侵蚀为传热内容的工程案例，设计“基础理论(一维稳态导热)→工程应用(炉缸炉底温度场分析)→深入探究(拐角部位温度分析)”的阶梯式问题链；以知识图谱关联理论节点与工程案例，构建系统化认知框架；依托超星学习通平台的知识图谱与课堂测评模块实现自动组卷与即时反馈，形成“问题激发–探究实践–优化”的教学闭环。两学期考核数据显示，课程平均成绩由68.2分升至73.4分(+7.6%)，优秀率由4.35%升至9.46%，增益主要发生在60分以上；不及格率由14.49%微增至14.86%，基本持平，提示该模式更接近提升型干预，而非面向学困生的补偿型干预。

Abstract: Traditional engineering education often faces challenges in connecting abstract theory with engineering practice and providing timely feedback. These challenges are particularly evident in core courses such as Metallurgical Transport Principles. Although students learn fundamental principles such as Fourier’s law, they may struggle to apply them to authentic engineering problems, including blast furnace hearth erosion. Drawing on cognitive apprenticeship and situated learning theory, this study proposes a teaching model integrating a progressive problem chain with AI-assisted tools. Blast furnace hearth erosion serves as an engineering case for the heat transfer component. The problem sequence progresses from one-dimensional steady-state heat conduction to temperature-field analysis of the hearth and, finally, two-dimensional heat transfer analysis near the hearth corner. A knowledge graph connects theoretical concepts with engineering applications, while the knowledge graph and classroom assessment modules of the Chaoxing Learning Tong platform support automatic quiz assembly and immediate formative feedback. Assessment data from two semesters show that the mean course score increased from 68.2 to 73.4 (7.6%), and the proportion of students attaining excellent grades rose from 4.35% to 9.46%. The gains were concentrated among students scoring at least 60, whereas the failure rate remained essentially unchanged (14.49% versus 14.86%). This pattern suggests that the model may support improvement among students with an adequate foundation, while offering limited remedial benefit to lower-performing students.