| 引用本文: | 施保华,井任月,杨 超,等.用于变电站保护压板状态识别的增强YOLO网络[J].电力系统保护与控制,2021,49(23):163-170.[点击复制] |
| SHI Baohua,JING Renyue,YANG Chao,et al.Enhanced YOLO network for status recognition of a substation protection plate[J].Power System Protection and Control,2021,49(23):163-170[点击复制] |
|
| 摘要: |
| 为解决变电站保护压板识别环境复杂、前景与背景难以分割和小目标检测困难等问题，提出了增强YOLO算法用于变电站保护压板的状态识别。首先，提出了局部残差聚合模块，对堆叠的残差模块的局部残差特征进行聚合，加强复杂环境下的识别能力。然后，将空域注意力机制嵌入到残差模块中，利用空间信息解决前景与背景难以分割的问题。最后，提出交叉空间金字塔模块用于提取全局上下文信息。利用标注的数据集进行验证，实验结果表明，增强YOLO算法相较于改进前算法，保护压板的状态识别效果显著提高。 |
| 关键词: 保护压板 状态识别 局部残差聚合 空域注意力 交叉空间金字塔 |
| DOI：DOI: 10.19783/j.cnki.pspc.210237 |
| 投稿时间：2021-03-05修订日期：2021-05-09 |
| 基金项目:国家自然科学基金项目资助(52077120)“大规模电力外送通道重合闸所致重大风险分析与规避控制策略研究”；国家自然科学基金项目资助(52007103)“计及热量迁移动态过程的电热耦合系统时空异构动态优化调度方法研究” |
|
| Enhanced YOLO network for status recognition of a substation protection plate |
| SHI Baohua,JING Renyue,YANG Chao,LI Zhenxing,LI Yanzhang |
| (1. College of Electrical Engineering and New Energy, China Three Gorges University, Yichang 443000, China;
2. State Grid Wuhan Power Supply Company, Wuhan 430000, China) |
| Abstract: |
| There are problems caused by the complicated recognition environment of a substation protection plate, the difficult segmentation of foreground and background, and the difficulty in detection of small targets. This paper proposes an enhanced YOLO algorithm for the state recognition of a substation protection plate. First, a local residual aggregation module is proposed to aggregate the local residual features of the stacked residual modules to strengthen the recognition ability in complex environments. Then, a spatial attention mechanism is embedded in the residual block and the spatial information is used to solve the problem that the foreground and background are difficult to segment. Finally, a cross-form space pyramid module is proposed to extract global context information. The labeled data set is used for verification. The experimental results show that the enhanced YOLO algorithm has significantly improved the status recognition effect of the protective plate compared with an improved algorithm.
This work is supported by the National Natural Science Foundation of China (No. 52077120) “Research on Major Risk Analysis and Control Strategy by Reclosing for Large-scale Power transmission Channel” and the National Natural Science Foundation of China (No. 52007103) “Research on Time-Space Heterogeneous Dynamic Optimal Scheduling Method of the Combined Electrical and Heating Systems Considering the Dynamic Process of Heat Transfer in Heat Network System”. |
| Key words: protection plate state recognition local residual aggregation spatial attention cross-form space pyramid |
