摘要
为有效协助变电站日常巡检工作,准确发现变电站异常情况,设计了基于视频AI(人工智能)识别的智能变电站机器人自动化巡视方法。采集变电站视频图像,利用二维Otsu阈值分割方法将视频图像中的前景目标与背景进行分离,最大程度排除变电站环境背景因素干扰,凸显出目标对象,并利用AI识别技术中的改进卷积神经网络对变电站目标图像进行特征提取与分类,获取智能变电站巡视结果。实验结果显示,该方法可以利用巡检机器人采集变电站巡检视频图像,经分割处理后识别到变电站异常情况,提示变电站管理人员进行及时管控。
To effectively assist in the daily inspection work of substations and accurately detect abnormal situations in substations,an intelligent substation robot automation inspection method based on video AI(artificial intelligence)recognition was designed.Collect substation video images,use two-dimensional Otsu threshold segmentation method to separate foreground targets and background in the video images,eliminate interference from substation environmental background factors to the greatest extent,highlight the target objects,and use improved convolutional neural networks in AI recognition technology to extract and classify features of substation target images,obtaining intelligent substation inspection results.The experimental results show that this method can use inspection robots to collect video images of substation inspections,and after segmentation processing,identify abnormal situations in the substation,prompting substation management personnel to control them in a timely manner.
作者
钟加勇
项波
刘丁豪
魏学备
程晓
ZHONG Jiayong;XIANG Bo;LIU Dinghao;WEI Xuebei(State Grid Chongqing Electric Power Research Institute,Chongqing 401123,China;Maicro(Nanjing)Technology Co.,Ltd.,Nanjing 211135,China)
出处
《自动化与仪表》
2024年第11期51-54,共4页
Automation & Instrumentation
基金
国网重庆市电力公司科技项目(2023渝电科技33#)。
关键词
视频图像
AI识别技术
智能变电站
机器人
自动化巡视
卷积神经网络
video images
AI recognition technology
intelligent substation
robot
automated inspection
convolutional neural network
