基于数字孪生与AI融合的铁塔基站三维数字资产构建  

Construction of 3D digital assets of tower base station based on digital twin and AI fusion

作　　者：王栋[1] 陈琦[1] 高阳 WANG Dong;CHEN Qi;GAO Yang(ZTE Coporation,Xi'an 710114,China)

机构地区：[1]中兴通讯股份有限公司,陕西西安710114

出　　处：《通信与信息技术》2025年第2期115-120,共6页Communication & Information Technology

摘　　要：通信基站是国家新基建、新质生产力的主要代表之一,截至2024年底,我国移动通信基站总数已达1265万个,其中5G通信基站总数达425.1万个。然而,通信基站中铁塔基站的资产管理仍然处于人工勘察采集、统计、维护的传统生产方式,采集效率、安全以及数据准确性一直是无线网络资产管理维护的痛点。随着数字孪生与人工智能技术的发展与融合,赋能铁塔基站的资产管理正逐步迈向数智化的增质提效。本研究通过引入无人机自动飞控图像采集、孪生实景建模、通信设备AI图像识别、三维单体化重建与体框拟合等关键技术,形成采集提效、建模增质、应用更具价值的铁塔基站三维数字资产的构建方法与实现。其中,通信设备AI图像识别是利用先进的目标检测算法和实例分割模型SAM,从二维高清图像中准确提取铁塔基站的设备图像掩码,结合孪生实景模型构建出铁塔基站的三维数字资产模型,从而实现诸如高精度工参测量、远程数字勘察设计、自动化设备巡检等多个业务场景的高价值应用。Communication base stations are one of the main representatives of the country′s new infrastructure and new quality pro⁃ductivity.By the end of 2024,the total number of mobile communication base stations in my country has reached 12.65 million,of which the total number of 5G communication base stations has reached 4.251 million.However,the asset management of tower base stations in communication base stations is still based on the traditional production method of manual survey collection,statistics,and maintenance.Collection efficiency,security,and data accuracy have always been pain points in wireless network asset management and maintenance.With the development and integration of digital twin(DT)and artificial intelligence(AI)technology,the asset management of tower base stations is gradually moving towards digital intelligence to increase quality and efficiency.This research introduces key technologies such as UAV automatic flight control image acquisition,twin reality modeling,communication equipment AI image recognition,three-dimen⁃sional single reconstruction and body frame fitting,etc.,to improve acquisition ef-ficiency,improve modeling quality,and application Construction method and implementation of more valuable tower base station three-dimensional digital assets.Among them,communica⁃tion equipment AI image recognition uses the advanced target detection algorithm and instance segmentation model SAM to accurately ex⁃tract the equipment image mask of the tower base station from the two-dimensional high-definition image,and combines it with the twin real-life model to build a three-dimensional digital asset model of the tower base station,thereby realizing high-value appli-cations in multiple business scenarios such as high-precision industrial parameter measurement,remote digital survey and design,automated equip⁃ment inspection,etc.

关 键 词：数字站点孪生 图像模型训推 实景三维重建 设备体框拟合 

分 类 号：TN92[电子电信—通信与信息系统]