基于不确定性分析的给水管网压力-漏损响应机制及组合建模技术

Based on the data of the leakage condition of water supply network in both domestic and international, the leakage state with different pipe types, environmental factors, leakage forms and area will be simulated by experimental study and calculation. The pressure - leakage relation model and the parameter range corresponding table under different conditions are established based on the improved FAVAD formula, though the research of the leakage influencing factors and their synergistic effects under different environmental conditions with various pipe materials. The simulation and analysis of unsteady leakage characteristics as well as the CFD simulation of pipeline flow pattern and flow process under different conditions are carried out to optimizing and adjusting the leakage model. Thereafter, the variation pattern base of the leakage pressure wave under unsteady flow state is established. According to the leakage event with random attribute,the sampling simulation based on the Latin Hypercube (LH) method with uncertainty analysis is carried out to simulate the expected influence value of the leakage event on hydraulic function of water supply network. The hydraulic characteristics and the influencing factors of water supply network is analyzed, then the response mechanism of leakage to the working condition is analyzed to establish the combined micro-model of water supply network taking the pressure-leakage relationship into account. According to the nonlinear characteristics of the combined micro-model, the effectiveness of the solution will be realized through the model conversion and inversion optimization algorithm, which will provide the technical support for the leakage control and operation management of water supply network.

在收集国内外供水管网漏损工况资料的基础上，通过实验研究和计算模拟相结合，开展不同管道类型、不同环境因素、不同漏损形态及面积的漏损状态模拟；研究各管材类型不同环境条件下的漏损影响因子及其协同作用，在改进FAVAD公式的基础上，建立不同状态下的压力-漏损关系模型和参数范围对应表。开展非稳态漏损特性分析与模拟；并对不同工况下的管道流动状态和流动过程进行CFD仿真，优化和校正漏损模型，构建非稳定流漏损压力波变化模式库；以漏损事件的随机性特点为切入点，采用基于不确定性分析的拉丁超立方（Latin Hypercube，LH）法抽样模拟，分析漏损事件发生对管网水力功能影响的期望值；分析供水管网的水力特性和影响因素，剖析漏损与管网工况的响应机制，建立引入压力-漏损关系的管网组合微观模型，针对管网组合模型的非线性特征，通过模型转换和反演优化等算法，实现求解的有效性，为供水管网漏损控制和运行管理提供技术支持

在大量搜集城市供水管网漏损现状及漏损历史记录的基础上，构建了管网属性及图形数据库，进而建立基于随机森林算法的管网漏失可能性模型进行漏失可能性评价。采用CFD-Fluent软件进行管网流动状态模拟仿真及漏失工况分析，在对水压-漏失量关系模型进行模拟分析和工况试验的基础上，建立不同漏损状态下的压力-漏损关系模型和参数范围，并开展非稳态漏损特性分析与模拟，基于模拟退火算法（SAA）和支持向量机（SVM）构建供水管网漏损特征预测模型，通过仿真模拟试验平台进行非稳态多工况训练学习，进而开展漏口形状、漏失量等漏损特征预测实验，对铸铁管、PVC、PE等多管材的漏失工况预测效果良好。通过对供水管网漏失信号源特征分析，建立基于Fast ICA算法的漏失量估算模型，通过模型分析与信号还原，实现供水管网的漏失量估算。以开源EPANET-TOOLKIT为基础进行二次开发，通过算法改进和功能扩展，独立研发了“供水管网分析模拟平台”，开发的“供水管网节点流量自适应动态优化技术”已获批发明专利，“一种开放式局域供水管网动态自适应仿真建模方法”已公开并在审查中。在前期研究的压力-漏损响应关系模型的基础上，建立基于压力驱动的供水管网漏失水力组合模型，随着管网压力的变化动态校正节点漏失水量分配，使供水管网水力模型更好地与实际节点水量分配相吻合，并通过优化算法确定泄漏系数c，研发了基于压力驱动的供水管网漏失组合建模技术，进而建立基于压力驱动漏失水力模型的漏损定位反问题模型，并将技术应用于漏失定位的实践探索中。.针对供水管网运行与漏失控制中压力监测点的重要作用，开展压力监控点的优化布置研究，并在漏失水力模型和压力在线监控的基础上建立管网漏损定位模型，并采用智能群体优化算法进行求解与比较，在北方城镇选择城镇工业区进行实例验证，具有较高的准确性和稳定性，研究成果可为智慧供水建设提供有力的技术支撑。