首页|期刊导航|现代教育技术|多模态数据融合驱动高校工程训练数智化转型的技术路径

多模态数据融合驱动高校工程训练数智化转型的技术路径OA

Technical Pathway for the Digital-Intelligent Transformation of University Engineering Training Driven by Multimodal Data Fusion

中文摘要英文摘要

工程实践教育是培养工程实践能力的重要环节.作为工程实践教育的重要手段,工程训练当前存在场景数据表征融合、模式革新和生态重构方面的多重问题,与工程人才培养需求的适应性、多样性、开放性尚不完全匹配.为此,文章聚焦工程训练数智化转型的关键问题,即如何利用数智技术开展多模态数据融合与分析、赋能工程实践教学过程与评价模式创新,首先梳理工程训练数字化、智能化转型的核心需求;其次,构建"硬件基础设施—数据底座层—多模态数据融合—场景应用—数据可视化"的数智化转型架构;最后,分析由"知识图谱—技能图谱"向数据观测点的映射、"数据感知—智能融合—认知赋能"的多模态数据融合、"效果评估—路径设计—数据驱动决策"场景功能应用的技术路线,并基于铸造课程展示了所提架构在多模态数据融合、教学过程赋能和学习效果综合评价的可行性.文章建立的工程实践教学闭环优化机制,为工程训练数智化转型提供了概念框架与案例参考.

Engineering practice education constitutes a critical component for cultivating engineering practical capabilities.As a major vehicle of engineering practice education,current engineering training is confronted with multiple challenges in the integration of scenario data representation and fusion,paradigm innovation and ecosystem reconstruction.It cannot fully meet the requirements of adaptability,diversity and openness required for engineering talent cultivation.Targeting the core challenge in the digital-intelligent transformation of engineering training,namely how to leverage digital and intelligent technologies to conduct multimodal data fusion and analysis and empower innovations in the teaching processes and evaluation paradigm of engineering practice education,this paper firstly sorts out the core demands of the digital and intelligent transformation of engineering training.Secondly,it constructs a digital-intelligent transformation architecture consisting of"hardware infrastructure-data foundation layer-multimodal data fusion-scenario application-data visualization".Subsequently,it analyzes the technical roadmap for scenario functional applications,encompassing the mapping from"knowledge graph–skill graph"to data observation points,multimodal data fusion of"data sensing–intelligent fusion–cognitive empowerment",and scenario function applications of"effective evaluation–pathway design–data-driven decision-making".Taking a casting course as an example,this study demonstrates the feasibility of the proposed architecture in multimodal data fusion,teaching process empowerment and comprehensive evaluation of learning outcomes.The closed-loop optimization mechanism for engineering practice teaching established in this study provides a conceptual framework and case reference for the digital-intelligent transformation of engineering training.

杨明亮;杨建新;郑养波;徐伟国;季林红;李双寿

清华大学 基础工业训练中心,北京 100084||清华大学 机械工程系,北京 100084清华大学 基础工业训练中心,北京 100084清华大学 基础工业训练中心,北京 100084清华大学 基础工业训练中心,北京 100084清华大学 机械工程系,北京 100084清华大学 基础工业训练中心,北京 100084

社会科学

数智化转型工程训练多模态数据融合人工智能

digital-intelligent transformationengineering trainingmultimodal data fusionartificial intelligence

《现代教育技术》 2026 (9)

112-121,10

本文为清华大学实验室创新基金重点项目"工程训练数智化赋能技术研究"(项目编号:202515007)、北京市教育科学"十四五"规划2025年度优先关注课题"大中小衔接的人工智能素养培养体系研究"(项目编号:CEDA25012)的阶段性研究成果.

10.3969/j.issn.1009-8097.2026.09.012

评论