基于深度学习的冷冻站工业表计读数识别  

Deep Learning-Based Industrial Meter Reading Recognition for Chiller Plants

作　　者：黄达 王智永[1] 刘媛[1] 王有友 舒誉 段希希[1] Huang Da;Wang Zhiyong;Liu Yuan;Wang Youyou;Shu Yu;Duan Xixi(Power Administrative Support Department of China Academy of Space Technology,Beijing 100094,China)

机构地区：[1]中国空间技术研究院动力行政保障部,北京100094

出　　处：《绿色建造与智能建筑》2024年第8期105-110,共6页Green Construction and Intelligent Building

摘　　要：在冷冻站的AR巡检过程中,为了实现工业表计读数的自动识别并提高巡检效率,本文提出了一种基于深度学习的冷冻站工业表计读数识别算法。首先通过AR眼镜采集表计的图片,然后使用目标检测算法检测出图片中的表计,将检测到的表计裁剪出来并输入到图像分割算法中,接着将表盘上的指针和刻度分割出来,最后使用极坐标变换将弧形的刻度展开成直线,根据指针位置和量程确定最终的读数。该方法已应用在实际的冷冻站AR巡检中,效果良好.In the AR inspection process of the chiller plant,to achieve automatic recognition of industrial meter readings and improve inspection efficiency,this paper proposes a chiller plant industrial meter reading recognition algorithm based on deep learning.First,images of the meters are captured using AR glasses.Then,a target detection algorithm is used to identify the meters in the images.The detected meters are cropped and input into an image segmentation algorithm,which segments the pointers and scales on the dial.Finally,the arc-shaped scales are converted into straight lines using polar coordinate transformation,and the final reading is determined based on the pointer position and the measurement range.This method has beenapplied inpractical AR inspections of chiller plants with good results.

关 键 词：AR巡检 深度学习 表计识别 冷冻站 

分 类 号：TU855[建筑科学]