基于深度学习的P&ID图纸智能识别
陈彦 1王优优 1刘飞 1李康杰 1张小闻 1韩嘉航1
作者信息
- 1. 石化盈科信息技术有限责任公司,江苏南京 211100
- 折叠
摘要
在石油化工领域,图纸数字化是一种重要的技术趋势.P&ID工程图纸是一种常用的工艺流程图,包含了成套设备、管道、自控仪表等大量信息.探讨了如何构建一套基于人工智能的P&ID图纸智能识别方法,通过计算机视觉技术智能识别并提取图纸信息,包括图例符号、文字、线条和表格等,同时可以根据管线连接关系进行关联,以分析管线连通性.该方法可大幅提高图纸识别和分析效率,实现P&ID图纸的数字化转换、存储和管理.
Abstract
In the field of petrochemical industry,the digitization of drawings is an important technological trend.P&ID drawings are a kind of commonly used process flow diagrams,which contain a large amount of information about complete sets of equipment,pipelines,automatic control instruments and so on.This paper discusses how to construct a set of artifi-cial intelligence-based intelligent recognition method for P&ID drawings,which intelligently recognizes and extracts drawing information,including legend symbols,text,lines and tables,etc.,through computer vision technology,and at the same time,it can be correlated according to pipeline connectivity in order to analyze pipeline connectivity.The method in this paper can greatly improve the efficiency of drawing recognition and analysis,and realize the digital conversion,storage and management of P&ID drawings.
关键词
P&ID图纸/工程图纸识别/计算机视觉/卷积神经网络Key words
P&ID drawings/engineering drawings recognition/computer vision/convolutional neural network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据