Method for automatically identifying machine patrol image defects of power transmission line based on yolos

Abstract

The invention discloses a yolo-based power transmission line machine patrol image defect automatic recognition method. The method comprises the following steps of S1, shooting a machine patrol image of a power transmission line; S2, noise reduction processing is carried out on the machine patrol image; S3, marking the machine patrol image after the noise reduction processing; S4S4, integrating themarked machine patrol images into a sample set, and establishing a sample database; S5, training a yolol model by utilizing the data of the sample database; and S6, detecting the machine patrol imageby using the trained yolo model, and generating a detection report according to a detection result. Visible light images of key equipment of a power transmission line are collected in an unmanned aerial vehicle inspection mode; defects are automatically extracted by using a deep learning algorithm, and the defects of key equipment of the power transmission line are intelligently identified and positioned through sample training, so that the problems of high risk, large interpretation workload and high error rate of manual inspection are effectively solved, reference is effectively provided for electric power overhaul, and safe and stable operation of a power grid is ensured.

Description

technical field [0001] The invention belongs to the field of transmission line detection, and in particular relates to a yolo-based automatic identification method for defects in machine patrol images of power transmission lines, which can be used for detecting and locating defects in machine patrol images of power system transmission lines. Background technique [0002] In modern society, electricity is an inseparable source of energy in daily life. The power system has become one of the most important material foundations supporting modern social civilization. The development level of the power industry has become an important symbol reflecting the country's economic development. Transmission lines are an important part of the power system. Due to their long-term exposure to the natural environment, they must not only bear the internal pressure of normal mechanical loads and electrical loads, but also withstand pollution, lightning strikes, strong winds, floods, landslides...

AI Technical Summary

Problems solved by technology

[0003] However, due to the extremely wide coverage of power lines, and through the mountains and vast forests, even some transmission lines need to cross no-man's land, traffic dead zone and communication blind zone, resulting in the operation and maintenance of transmission lines and towers and sudden failures It is extremely difficult to find. The traditional manual line inspection method can no longer meet the operation and maintenance requirements of EHV and UHV power grids. The inspection work is not only intensive but also highly dangerous. Therefore, transmission line inspection has gradually evolved into "human inspection" + " machine inspection mode

However, due to the complex operating conditions of the power transmission channels during the helicopter or drone inspection process, the operation and maintenance personnel can obtain limited information on the site to judge the operation status of the equipment and channel conditions, and often cannot judge whether there is a defect in the transmission line in real time. The inspection image is sent to the server for manual interpretation by professional maintenance personnel

According to data, the image and video data generated by machine patrol operations in a southern province is about 40T each year. Manual interpretation of these machine patrol images is not only a huge workload, it is impossible to provide guidance for the standardized process of line machine patrol operations, and it may cause unsatisfactory inspections. In place, missing important inspection items, etc., and manual interpretation can only complete a small amount of obviously defective data interpretation, resulting in a large number of machine inspection images that can only be stored in the hard disk and cannot be used

[0004] Therefore, in order to solve the problems of manual interpretation of machine patrol images, such as large amount, incompleteness, lack of important inspection items, and untimely data analysis, etc., a method for deep feature mining and utilization of a large amount of machine patrol image data can be invented, which can effectively detect power transmission. The method of main defects of transmission lines such as line appearance, operating environment, and component abnormalities is particularly important, which can effectively provide reference for power maintenance and ensure safe and stable operation of the power grid

Images