大学化学 >> 2026, Vol. 41 >> Issue (7): 430-440.doi: 10.12461/PKU.DXHX202505070

师生笔谈 上一篇    

卤代正丁烷制备实验中的红外光谱AI智能识别和教学辅助系统的开发

畅正轩, 蒋浩阳, 赵卫光   

  1. 南开大学化学学院, 天津 300371
  • 收稿日期:2025-05-25 录用日期:2025-08-01 发布日期:2026-06-27
  • 通讯作者: 赵卫光 E-mail:zwg@nankai.edu.cn Weiguang Zhao

Development of an AI-powered infrared spectroscopy recognition and teaching assistance system for halogenated n-butane preparation experiment

Zhengxuan Chang, Haoyang Jiang, Weiguang Zhao   

  1. College of Chemistry, Nankai University, Tianjin 300071, China
  • Received:2025-05-25 Accepted:2025-08-01 Published:2026-06-27
  • Contact: Weiguang Zhao E-mail:zwg@nankai.edu.cn

PDF

151

赞

摘要： 随着人工智能技术的快速发展，其在化学教育中的应用已成为教育变革的核心驱动力。本文以卤代正丁烷合成实验为例，基于南开大学AI创新平台(NK-GeniOS)，构建了红外光谱智能识别与教学辅助系统。通过单调性抗锯齿优化法实现了对于图片形式的红外谱图中产品及杂质红外吸收峰的快速准确识别，通过设计分析Agent与评价Agent的多智能体协同机制，系统实现了对实验数据的精准解析与个性化反馈，解决了传统教学中谱图解读依赖教师经验、缺乏实时指导的痛点。此外，通过组建项目小组，以“研究-汇报-讨论”的模式实现了实验教学的分层赋能，不同层次的学生都能有所收获，并极大地激发了学生的科研兴趣，79%的同学开始愿意参与到类似的研究中，参与翻转课堂的意愿也从20%提高到71%。

关键词: 红外光谱, 人工智能, 有机化学实验, 跨学科融合

Abstract: The rapid advancement of artificial intelligence technology has positioned it as a pivotal driver in transforming chemical education. This study presents an infrared spectroscopy intelligent recognition and teaching assistance system developed for n-butyl halide synthesis experiments, based on Nankai University’s AI innovation platform (NK-GeniOS). The system employs monotonicity anti-aliasing optimization to achieve rapid and accurate identification of infrared absorption peaks from spectral images for both products and impurities. Through a multi-agent collaborative framework comprising analysis Agent and evaluation Agent, the system provides precise experimental data interpretation and personalized feedback, addressing the limitations of traditional teaching methods that heavily rely on instructor experience and lack real-time guidance. Implemented via project-based learning groups following a “research-presentation-discussion” model, this approach enables tiered learning outcomes for students at different levels. The system significantly enhanced student engagement, with 79% of participants expressing willingness to participate in similar research projects, while flipped classroom participation rates increased dramatically from 20% to 71%.

Key words: Infrared spectroscopy, Artificial intelligence, Organic chemistry experiment, Interdisciplinary integration