网络首发

基于机器视觉的预制叠合板智能检测关键技术

于海洋¹’²,李海生¹,彭伟³,孙宏伟¹,于晓光²

作者简介:

于海洋(1991.11),男,山东青岛人,荣华建设集团有限公司研发工程师。E-mail: 809086549@qq.com

作者单位:

1. 荣华建设集团有限公司,青岛 266000;2. 荣华(青岛)建设科技有限公司,青岛 2660003. 山东建筑大学 济南250000

基金项目:

2021年山东省重点研发计划(重大科技创新工程)-“绿色智能建造和建筑工业化关键技术与成套装备”,项目编号:2021CXGC011205;2022年山东省住房城乡建设科技计划项目(项目编号:2022-K7-6)

摘要:

装配式建筑采用工业化生产、装配化施工,具有施工效率高、节能环保等特点,受到了国家的大力扶持,而预制混凝土叠合板作为预制率、装配率非常高的预制构件,其质量优劣关系着整个工程。预制叠合板在生产过程中存在着一些问题,如混凝土浇筑之前的隐蔽验收工作采用人工检查的方法,人工投入量大,质量检测水平和效率低下,容易漏检、错检,严重制约了混凝土叠合板的生产效率。本文以预制混凝土叠合板为研究对象,开展了图像畸变校正算法和数据增强算法研究,开展了改进YOLOv7预制叠合板精准识别算法和钢筋排布和间距识别方法研究,开发了一套基于机器视觉的预制叠合板智能检测控制系统,系统可对叠合板尺寸误差、预埋件位置及数量进行智能评估,从而减少质检人员投入,提高生产效率和产品合格率。

English:

Prefabricated buildings adopt industrial production and prefabricated construction, which have the characteristics of high construction efficiency, energy conservation and environmental protection, and have received strong support from the country. As prefabricated components with very high prefabrication and assembly rates, the quality of prefabricated concrete composite panels is related to the entire project. There are some problems in the production process of prefabricated composite panels, such as the use of manual inspection for concealed acceptance work before concrete pouring, which results in a large amount of manual input, low quality inspection level and efficiency, and easy omission and misinspection, seriously restricting the production efficiency of concrete composite panels. This article takes prefabricated concrete composite panels as the research object, conducts research on image distortion correction algorithm and data enhancement algorithm, improves YOLOv7 prefabricated composite panel precise recognition algorithm and steel bar layout and spacing recognition method, and develops a machine vision based intelligent detection and control system for prefabricated composite panels. The system can intelligently evaluate the size error, embedded part position and quantity of composite panels, thereby reducing the investment of quality inspectors, improving production efficiency and product qualification rate.