基于深度学习叩诊方法的钢筋混凝土构件损伤识别

打开文本图片集

中图分类号：TU375 文献标志码：A DOI： 10.15935/j.cnki.jggcs.202504.0014

AbstractReinforced concrete components play a crucial role in modern construction projects，and their damage statesare closely related to the overall safety of buildings.Traditional damage detection methods mainly rely on manual experience for judgment，facing the dilemma of insuffcient accuracy and timeliness in detecting structural damage. This study proposes an innovative percussion method based on deep learning technology，which achieves automated and precise identification of the damage states of reinforced concrete components by integrating acoustic signals with deep learning algorithms.Taking reinforced concrete shear walls as the research object，this paper constructs an acoustic dataset of shear wals covering various damage states，uses a convolutional neural network （CNN） to extract acoustic features，and then employs a classification algorithm to determine damage states.Experimental verification shows that the proposed deep learning-based percussion method exhibits excellent accuracy and eficiency indamage identification of shear wall components，demonstrating outstanding application potential and engineering application prospects.It providesan effcient and non-destructivenew solution for the field of damage identification of reinforced concrete components.

Keywordsdeep learning，percusson method，damage identification，convolutional neural network， reinforced concrete shear wall

0 引言

我国钢筋混凝土结构量大面广，快速、准确、无损地识别在役钢筋混凝土结构的损伤状态一直是工程师和研究者们关注的重要问题。（剩余10957字）