基于人工神经网络的智能配网实时故障图像识别系统
A Real-Time Fault Image Recognition System of Intelligent Distribution Network Based on Artificial Neural Network
林园敏 1王鑫 1黄金钊 1陈诗韵1
作者信息
- 1. 广东电网有限责任公司广州番禺供电局,广东广州 510000
- 折叠
摘要
配网巡检的故障图像识别由于图像特征提取效果差,存在故障识别准确性低的问题,在该背景下,设计一种基于人工神经网络的智能配网实时故障图像识别系统.该系统根据模型-视图-控制器(model view controller,MVC)设计模式设计框架结构,包括无人机采集层、数据传输层以及地面站系统操作层.在硬件设计部分,借助无人机搭载拍摄设备,采集配电网设备图像,并通过无线数据传输电台将图像传输到地面站,实施图像处理与识别;在软件部分,利用人工神经网络算法,结合粒子群算法等,设计巡检路径规划、图像采集传输、图像特征提取以及图像识别 4 个关键功能模块.实验结果表明:在设计系统的应用下,识别出研究区域发生的配网故障;通过图像样本,系统识别结果准确性达到94.5%以上,证明了系统的有效性.
Abstract
The fault recognition accuracy of distribution network inspection is often low due to the poor extraction effect of image features.For this reason,an intelligent distribution network real-time fault image recognition system based on artificial neural network is designed.The system designs the frame structure according to MVC design pattern,which includes the UAV acquisition layer,data transmission layer and ground station system operation layer.In the hardware design part,with the help of the UAV equipped with photographing equipment,the image of distribution network equipment is collected,and the image is transmitted to the ground station through wireless data transmission station to implement image processing and recognition.In the software part,using artificial neural network algorithms combined with particle swarm optimization,four key functional modules are designed,including patrol path planning,image acquisition and transmission,image feature extraction and image recognition.The results show that under the application of the designed system,the distribution network faults in the research area are identified.In addition,through the image samples,the accuracy of the system identification results is more than 94.5%,which proves the effectiveness of the system.
关键词
人工神经网络/智能配网/巡检路径/图像识别Key words
artificial neural network/intelligent distribution network/inspection path/image recognition引用本文复制引用
基金项目
中国南方电网有限责任公司科技项目(082700KK52200001)
出版年
2024
NETL
NSTL
万方数据