Real-time Analytics · Risk Management
Yunnan Jiaotou needed a transport-domain AI system that could handle complex mountain highway construction, maintenance, operations, and safety workflows, but no mature transportation large model, no high-quality domain dataset, and no multi-model orchestration approach existed for the sector.
Reported impact
The program put more than 30 AI applications into production, delivered highway traffic forecasts with more than 90% accuracy, reduced congestion duration, and materially improved the speed of construction and safety inspection workflows.
Yunnan Communications Investment and Construction Group built the 'Green Beautiful Corridor' transportation large model on Huawei Cloud Stack to support highway construction, operations, maintenance, safety, and office work. The platform uses a cloud-edge-end architecture with one data center, one disaster-recovery center, and 35 edge nodes. Huawei also supplied an AI toolchain for data engineering, model engineering, agent development, and O&M, plus transport datasets with over 600 GB of industry corpus and 400 GB of enterprise knowledge. On that base, the team created four model families for industry understanding, traffic-state sensing and forecasting, multimodal perception, and decision generation and optimization. Those models power 30+ applications including traffic forecasting from toll, gantry, road, event, and weather data, as well as BIM review, drone inspection, video-based unsafe-behavior detection, and AI emergency-duty workflows.
Traffic flows, speeds, incidents, weather feeds, toll-station and gantry data, BIM drawings, drone imagery, and site video are collected into the cloud-edge-end platform and high-quality transport datasets. The large-model stack trains and serves forecasting and perception models, which then feed traffic-control decisions, construction review, safety alerts, and emergency-duty workflows.
"In the Mouyuan expressway project, inspection time fell from 3 hours to 20 minutes and efficiency improved by 65%. (original: 在牟元高速建设项目中，巡查效率由传统人工巡查3小时降低到无人机智能巡查20分钟，效率提升65%。)"
"The architecture implements a '1 data center + 1 disaster-recovery center + 35 edge computing nodes' cloud-edge-end intelligent computing base. (original: 实现“1数据中心+1容灾中心+35个边缘计算节点”的“云、边、端”算力基础设施架构，建成绿色低碳、安全可信的智算中心)"
"The project built more than 600 GB of industry corpus and more than 400 GB of enterprise knowledge to form a high-quality transport training and evaluation dataset. (original: 融合行业语料集，搭建超过600GB的行业知识语料和超过400GB企业专有知识库，形成公路交通行业首例高质量训练数据集与大模型评测数据集)"
"The traffic forecasting model uses data from 800+ gantries and 300+ toll stations, achieves accuracy above 90%, predicts 15 minutes ahead, and projects congestion 30 minutes into the future. (original: 实现对云南交投全路网800多门架、300多收费站实时流量小时级预测，准确率大于90%，可提前15分钟精准预测交通流量，为道路管控提供准确依据，精准评估事件（事故/施工）造成的拥堵程度，推演未来30分钟交通拥堵情况。)"
"Yunnan Jiaotou built the 'Green Beautiful Corridor · Transportation Large Model' on Huawei Cloud Stack and has already put more than 30 AI applications into use across the transport lifecycle. (original: 云南交投基于华为云Stack打造“绿美通道·交通大模型”，使能交通行业“建设、管理、养护、运营、服务、安全、办公”七大业务领域，实现全周期智能化，已落地30多个AI应用。)"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Huawei Cloud Sample Guangdong Joint Electronic Toll Service Co., Ltd. (广东联合电子服务股份有限公司) Joint Electronic Toll Service needed to govern massive highway transaction and mobility data, support more than 25 million ETC users, and turn that data into AI-assisted operational decisions for tolling, traffic… built 6 data-product lines and 63 data products Data Lake Insight (DLI) · DataArts Studio · ModelArts Huawei Cloud Sample Archived · Jan 15, 2026 Conch Group Conch needed to scale AI across cement production, but traditional model development was slow, hard to generalize, and too weak to support real-time quality, energy, equipment, and safety optimization at industrial… 28 equipment-monitoring scenarios AI training center · Huawei Cloud Stack · NLP large model Huawei Cloud Sample Archived · Sep 8, 2025 Ordos City Ordos needed a repeatable way to modernize hundreds of fragmented coal mines, but operators lacked shared standards, interoperable data, scalable AI infrastructure, and reusable models for safety, efficiency, and… 5 demonstration mines launched Huawei Cloud Stack · MaaS application marketplace · Pangu mining large model 