李福龙,马孝栋,魏春雷,王天慧,陈洲生



[摘 要]新能源发电行业正朝着无人值守、少人值守的方向发展,变电站智能机器人巡检方案通过整合不同的技术来实现这一目标,从而替代运维人员执行对站内室外高压设备的巡视检查、对电表数值的现场读取、对电力设备运行数据的实时采集等任务,并预测可能发生的故障,旨在为巡检机器人提供电力设备自主识别能力,模拟人工巡检中对各种设备进行定制化检查,涉及的设备类型有变壓器、断路器、互感器、隔离开关、仪表、阀门等。鉴于变电站现场环境的复杂性,提出了一种基于深度学习的实时设备检测模型。结果表明,该模型的平均精度在90%以上,验证了该设备检测模型的实用性,并展示了其在巡检机器人系统上的应用潜力。
[关键词]电力巡检;机器人;目标检测;设备识别
[中图分类号]TM50 [文献标志码]A [文章编号]2095–6487(2022)02–0–04
Research on the Equipment Detection Model of New
Energy Substation Based on Deep Learning
Li Fu-long,Ma Xiao-dong,Wei Chun-lei,Wang Tian-hui,Chen Zhou-sheng
[Abstract]The new energy power generation industry is developing in the direction of unattended and few people on duty. The intelligent robot inspection solution for substations achieves this goal by integrating different technologies, thereby replacing operation and maintenance personnel to perform inspections and inspections of high-voltage equipment inside and outside the station, on-site reading of meter values, real-time collection of power equipment operating data and other tasks, and prediction of possible failures. This article aims to provide the inspection robot with the ability to independently identify power equipment, and to simulate the customized inspection of various equipment during manual inspections. The types of equipment involved in this research include transformers, circuit breakers, transformers, isolating switches, meters, and valves. In view of the complexity of the substation site environment, we propose a real-time equipment detection model based on deep learning. The results show the remarkable performance of the model, with an average accuracy of more than 90%. This work verifies the practicability of the equipment detection model and demonstrates its application potential in the inspection robot system.
[Keywords]power inspection;robot;target detection;equipment recognition
现阶段新能源变电站巡检主要是以人工巡检为主,部分设备的自动监测为辅。为确保站内设备安全稳定运行,站内会设置日常巡视检查制度,运维人员通过定期巡视检查关键设备的有无异常情况,从而判断设备是否正常运行,及时掌握生产运行中的异常情况,并采取相应处置措施,对于降低电力事故的发生概率和严重程度具有非常重大的意义。因此,变电站需要参考变电设备的实际运行工况,同时结合过往处置设备异常、故障和缺陷的教训和经验,制订详尽的检查方案。……