5G base station inspection system based on artificial intelligence technology

0 0
  • 0 Collaborators

This research includes professional UAV flight operation platform, artificial intelligence image processing system, GPU image deep learning, automatic object recognition and other characteristics, which can automatically identify the operation status and abnormal conditions of tower, base station . ...learn more

Project status: Concept

Mobile, Artificial Intelligence

Intel Technologies
DevCloud, DPC++

Overview / Usage

  1. Research and selection of UAV flight platform. For the demand of intelligent inspection, UAV flight platform needs to have the characteristics of hovering, good low-speed performance, small turning radius and so on. Therefore, rotor aircraft is more suitable than fixed wing aircraft. Compared with helicopter, multi rotor UAV has the advantages of simple structure, convenient operation, high reliability, low cost and low difficulty of logistics maintenance, which is more suitable for 5g base station inspection. At the same time, the process of UAV 5g base station inspection should be studied, including technical personnel requirements, inspection airspace application, flight conditions, etc.

  1. Base station defect labeling and data analysis based on panda module. In the routine inspection of 5g base station, the common defects include connector, label, wiring, fixing, lightning protection, engineering reference, equipment, tower body, foreign matter nine categories and 56 sub categories. The above defect images are collected and analyzed by using the panda module in Python. Based on the analysis of a large number of image data, the data cleaning is completed, and the defect image database is established, which establishes the foundation for the follow-up deep learning and training of intelligent image recognition.

  1. Artificial intelligence image recognition technology based on tensorflow. Tensorflow is a system that transmits complex data structure to artificial intelligence neural network for analysis and processing. It can be used in many machine deep learning fields such as speech recognition or image recognition. Based on the established defect image database, deep learning and training of image recognition are carried out. The image recognition program completed more than 9900 training of defect image neural network, then tested the defect image recognition, and the recognition accuracy should reach above 0.99.

  1. Product practicability verification and application field mining research, product testing is carried out in Jinan and Weifang first, testing more than 100 5g base stations, evaluating the inspection effect of the product, focusing on the independent identification of 5g base station defects. The application of the research products is demonstrated in the fields similar to 5g base station inspection, such as urban management, land and resources survey, power tower inspection, road lighting tower inspection, etc., so as to maximize the potential of the product application.
Comments (0)