摘要
在输电线路上,螺栓松动和导线破损是普遍存在的输电线路故障,检测出这些故障对电力系统的安全运行具有非常重要的意义。YOLO V3是一种准确率高实时性强的目标检测算法,因此提出了基于深度卷积神经网络的YOLO V3目标检测方法,识别和定位输电线路上的螺栓和破损导线。通过对YOLO V3算法进行适应性的改进,使得输电线路上的检测任务更加快速和准确。利用TensorFlow框架搭建目标检测网络,训练得到最终检测模型并测试。实验结果表明,该输电线路故障检测方法实时性强、准确率高,能够满足自动检测输电线路上的螺栓和破损导线的要求,极大地提高了电力系统检修工作的"智能化"。
On the transmission line, loose bolts and broken wires are common transmission line faults, and detecting these faults is of great significance to the safe operation of the power system.YOLO V3 is an object detection algorithm with high accuracy and high real-time performance,therefore, YOLO V3 object detection method based on deep convolutional neural network is proposed to identify and locate the bolts and damaged wires on the transmission line. By improving the adaptability of the YOLO V3 algorithm, the detection task on the transmission line is faster and more accurate. The TensorFlow framework is used to build the target detection network, and the final detection model is obtained and tested through training.The experimental results show that the method can meet the requirements of automatic detection of bolts and broken wires on transmission lines with strong real-time performance and high accuracy, and innovatively improve the "intelligence" of power system maintenance.
作者
张迪
樊绍胜
ZHANHG Di;FAN Shao-sheng(School of electrical and information engineering Changsha University of Science & Technology,Changsha 410114 China)
出处
《自动化技术与应用》
2019年第7期125-129,共5页
Techniques of Automation and Applications
关键词
输电线路
卷积神经网络
故障检测
YOLO
V3
transmission line
convolutional neural networks
fault detection
YOLO V3