[1]谷俊鹏,刘静,黄俊,等.数据+模型“双驱动”的排水管网风险评估与预测体系[J].中国给水排水,2025,41(20):17-24.
GUJun-peng,LIUJing,HUANGJun,et al.Risk Assessment and Prediction System for Drainage Networks Driven by Data and Models[J].China Water & Wastewater,2025,41(20):17-24.
点击复制
GUJun-peng,LIUJing,HUANGJun,et al.Risk Assessment and Prediction System for Drainage Networks Driven by Data and Models[J].China Water & Wastewater,2025,41(20):17-24.
数据+模型“双驱动”的排水管网风险评估与预测体系
中国给水排水[ISSN:1000-4062/CN:12-1073/TU]
卷:
第41卷
期数:
2025年第20期
页码:
17-24
栏目:
出版日期:
2025-10-17
- Title:
- Risk Assessment and Prediction System for Drainage Networks Driven by Data and Models
- 关键词:
- 风险评估与预测; StormDesk; 极限梯度提升（XGBoost）; 长短期记忆网络（LSTM）
- Keywords:
- risk assessment and prediction; StormDesk; XGBoost; LSTM
- 摘要:
- 排水管网是城市基础设施的重要组成部分，其安全运行对保障城市可持续发展具有关键作用。随着我国城市化进程的加速推进，排水管网系统承受着日益增长的运行压力，对风险评估和预测提出了更高的要求。为提高排水管网运维管理水平，发掘新质生产力，充分开发利用数据资源，提升风险管理与治理能力，提出数据+模型“双驱动”的城市排水管网风险评估与预测体系。通过融合物联网、大数据分析、模型仿真等先进技术，收集整合管网参数、运行状态、缺陷情况以及周边环境等信息，构建并训练机器学习模型，准确识别影响管网风险的关键因素，对未知风险进行预测，对治理方案进行比选和评估。研究结果表明，该体系的应用实现了排水管网运行状态的实时监控和风险预测，低风险预测模型的准确率可达94.6%，为实施预防性维护和调整运行策略、降低风险发生概率提供了有力保障，为城市排水管网的优化管理提供了科学依据。
- Abstract:
- The drainage network is an important component of urban infrastructure, and its safe operation plays a crucial role in ensuring sustainable urban development. The rapid urbanization in China is placing increasing operational pressure on drainage network systems, thereby heightening the demand for effective risk assessment and prediction. In order to improve the operation and management level of drainage pipe networks, explore new quality productivity, fully develop and utilize data resources, enhance risk management and governance capabilities, a risk assessment and prediction system driven by data and models for urban drainage pipe networks is proposed. By integrating advanced technologies such as the Internet of Things, big data analysis, and model simulation, we collect and integrate information on pipeline parameters, operating status, defect conditions, and surrounding environment. Machine learning models are constructed and trained to accurately identify key factors influencing pipeline risks, predict potential risks, and compare and evaluate management strategies. The research results indicate that the application of this system has achieved real-time monitoring and risk prediction of the operational status of the drainage network. The accuracy of the low-risk prediction model can reach 94.6%, providing strong support for implementing preventive maintenance and adjusting operation strategies, reducing the probability of risk events, and informing the optimization of urban drainage networks.
更新日期/Last Update:
2025-10-17