Open AccessAn accurate and fast detection method for railway lightning protection component
Author(s) -
Suining Wu,
Bin Li,
Zhixin Wang,
Xiaoguang Chen,
Long Shi,
Shulin Tan
Publication year - 2020
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/792/1/012010
Subject(s) - component (thermodynamics) , lightning (connector) , lightning strike , computer science , fault (geology) , sorting , fault detection and isolation , engineering , reliability engineering , lightning arrester , artificial intelligence , electrical engineering , algorithm , actuator , power (physics) , geology , physics , quantum mechanics , seismology , thermodynamics
Lightning protection component is used to display the working condition of railway cable networks. Lightning strike damage cable networks, hence, the research of detection method for railway lightning protection is of great importance. So far, there are limited methods to solve this problem. Existing methods are based on electrical and magnetic signals. However, these method can not meet the acquirement of speed and accuracy. In this paper, we proposed a accurate and fast detection method for railway lightning protection component. The proposed method is based on deep neural network named W-YOLO. To solve the imbalanced samples problem of fault components and common components, we devised a novel classification loss function according to the proportion components. Since the raw class and location coordinate of components are lack of meaning, a sorting method of elements based on the angel and distance threshold is proposed. From the experiments, the proposed detection method is able to detect railway lightning protection component accurately in real time.