基于SSD目标检测的硬压板图像识别  

Improved SSD Algorithm Based on Pressure Plate Status Recognition

作　　者：卢志钢 洪乙铭 LU Zhi-gang;HONG Yi-ming(Shishi Power Supply Bureau of Fujian Power Grid,Shishi 362700,China)

机构地区：[1]国网石狮市供电公司,福建石狮362700

出　　处：《电气开关》2021年第4期37-39,42,共4页Electric Switchgear

摘　　要：为了减少变电站运维人员由于人为记录而导致硬压板状态误投退操作,本文提出了对传统SSD目标检测法进行改进,运用所改进的SSD网络模型对变电站中的硬压板图像进行识别,通过收集来的硬压板图片进行数据扩充、网络训练,得到能够识别硬压板状态的改进型SSD目标识别网络,最后通过验证集进行网路验证,结果证明了所提出的改进型SSD目标检测算法能够很好的识别硬压板状态,提高了变电站智能化水平。In order to reduce the statistical errors of the pressure plate status caused by the miswriting and omission of writing by the substation operation and maintenance personnel,this paper proposes an improved SSD algorithm to identify the pressure plate status images in the substation.Data expansion and network training are used to obtain an improved SSD network that can identify the state of the pressure plate.Finally,the network verification is performed through the verification set.The results prove that the proposed improved SSD target detection algorithm has a good performance in the identity of the state of the pressure plate and improve the intelligence of the substation level.

关 键 词：改进型SSD算法 DenseNet 目标检测 硬压板 

分 类 号：TM93[电气工程—电力电子与电力传动]