摘要
检测设备的更新为实现变电站检测仪表的自动识别提供了契机,但大多数基于图像识别的方法对图像质量都有严格的要求,因而不能直接应用到变电站检测仪表的识别。本文提出了一种基于深度学习的变电站表计智能识别方法。首先采用模糊局部信息C均值聚类方法实现不同场景下的表计图像分类。然后,使用FasterR-CNN方法检测目标表计表盘的位置,并根据检测框调整图像位置。最后,利用霍夫变换检测指针位置进而获得读数。识别方法的稳定性和准确性已得到实验结果验证,具有广阔的应用前景。
The update of testing equipment provides an opportunity to realize automatic identification of substation testing instruments.Most methods based on image recognition have strict requirements on image quality,so they cannot be directly applied to the recognition of substation detection instruments.This paper proposes an intelligent recognition method for substation meters based on deep learning.First,the fuzzy local information C-means clustering method is used to realize the classification of meter images in different scenarios.Then,use the Faster R-CNN method to detect the position of the target meter dial and adjust the image position according to the detection frame.Finally,the Hough transformation is used to detect the pointer position and obtain the reading.The experimental results verify the recognition method’s stability and accuracy,and it has broad application prospects.
作者
张世琦
苗俊杰
李峻宇
陈辰辰
龙彬
ZHANG Shiqi;MIAO Junjie;LI Junyu;CHEN Chenchen;LONG Bin(State Grid Hengshui Electric Power Supply Company,Hengshui 053000,China;State Grid Hebei Electrice Power Co.,Ltd.,Shijazbuang 050021,China;Xian Jiaotong University,Xian 710049,China;Wuhan Kangpu Evergreen Software Technology Co,Ltd.,Wuhan 430073,China)
出处
《河北电力技术》
2022年第4期1-5,65,共6页
Hebei Electric Power
关键词
深度学习
变电站
智能识别
检测仪表
deep learning
substation
intelligent identification
detection instrument
