基于Web的管廊可视化信息管理系统设计与实现  

Design and Implementation of Utility Tunnel Visual Information Management System Based on Web

作　　者：陈登峰[1] 张温 耿建勤 刘国 肖海燕[2] Chen Dengfeng;Zhang Wen;Geng Jianqin;Liu Guo;Xiao Haiyan(School of Building Equipment Science and Engineering,Xi'an University of Architecture and Technology,Xi'an710055,China;School of Information and Control Engineering,Xi'an University of Architecture and Technology,Xi'an710055,China)

机构地区：[1]西安建筑科技大学建筑设备科学与工程学院,西安710055 [2]西安建筑科技大学信息与控制工程学院,西安710055

出　　处：《计算机测量与控制》2020年第5期170-174,共5页Computer Measurement &Control

基　　金：国家自然科学基金项目(51705393);陕西省教育厅专项科研项目(14JK1408)。

摘　　要：城市化进程促进了城市地下综合管廊的飞速发展,但目前管廊的巡检等运维管理工作大多以人工为主,且信息化水平低下,而且管廊内的管线复杂、人工运维管理效率低且危险隐患高;文章提出一种基于Web的管廊建筑结构与建筑信息轻量可视化、CO环境数据可通过浏览器在线训练与预测的系统;通过Revit二次开发技术将建筑信息模型轻量化解耦并通过WebGL技术将几何模型加载在浏览器前端页面,后台服务程序可以通过通信线程池将终端节点采集到的环境数据存入Mysql数据库,并与浏览器前端动态耦合;通过整合Tensorflow.js可以在浏览器前端实现基于LSTM算法的CO浓度预测模型在线训练与预测;经过测试,模型加载平均耗时114.24 ms,单次交互平均耗时20 ms,预测精确度为86.3%,满足系统设计需求。Urbanization promoted the rapid development of utility tunnel.However,there are some problems in the management of utility tunnel currently such as management mean depends on manual,the complicated pipelines in utility tunnel,the low efficiency of manual management and the high risk factors.The article puts forward a web-based utility tunnel visualization information management system based on web to realize the lightweight visualization of building structure and building information,online prediction and model training of carbon monoxide(CO)environmental data in web browser.Management system uses the secondary development of Revit to decouple the BIM model and loads the geometric model in the web page by WebGL.The background services put the environment data which is collected by the terminal nodes to Mysql database through the communication thread pool,and dynamically couple with the browser web page.Integrating Tensorflow.js to implement online training and prediction of CO based on long short-term memory(LSTM)in the browser.According to the test,the loading process of model takes 114.24 ms on average,the average time of a single interaction is 20 ms,and the accuracy of prediction is 86.3%.The management system meets the design requirements.

关 键 词：综合管廊 WEB 轻量可视化 Tensorflow.js 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]