基于SAA-SVM模型的供水管网漏损预测技术研究  

Research on Leakage Prediction Technology of Water Supply Network Based on SAA-SVM Model

作　　者：张振星 吕谋[1] 赵桓 ZHANG Zhen-xing;LV Mou;ZHAO Huan(School of Environmental and Municipal Engineering,Qingdao University of Technology,Qingdao 266033,China)

机构地区：[1]青岛理工大学环境与市政工程学院,山东青岛266033

出　　处：《水电能源科学》2022年第11期137-140,共4页Water Resources and Power

基　　金：国家自然科学基金项目(51778307);山东省重点研发计划项目(2019GSF111003)。

摘　　要：为探究城镇供水管网漏失量、漏口失效形式与漏失压力之间的关系,根据供水管网的水力特性和漏损变化规律,基于模拟退火算法(SAA)的突跳特性和支持向量机(SVM)的稀疏性和稳健性,构建了SAA-SVM组合模型,利用SAA对SVM的参数c、g进行优化,并通过供水管网仿真模拟试验平台将多工况实时监测漏失数据作为样本数据集输入至SAA-SVM模型,经SAA-SVM对样本数据集进行处理、训练,进而展开漏损预测。结果表明,支持向量机(SVM)的准确度和跳出局部最优解的能力大幅提高,优化后的SAA-SVM能对漏口失效形式、漏失量进行快速准确的预测,满足供水管网多工况漏损预测的需求。In order to explore the relationship between leakage amount, leakage failure forms and leakage pressure of urban water supply networks, the SAA-SVM combination model was constructed according to the hydraulic characteristics and leakage change law of water supply network. Simulated annealing algorithm(SAA) could optimize the parameters c and g of support vector machine(SVM) based on the jump characteristics of SAA and the sparsity and robustness of SVM. Through the simulation experiment platform of water supply pipe network, the multi-condition real-time monitoring leakage data were input as sample data set of the SAA-SVM model. The SAA-SVM was used to process and train the sample data set, and then the leakage prediction was obtained. The results show that the accuracy of the SVM and the ability to jump out of the local optimal solution are greatly improved. The optimized SAA-SVM can quickly and accurately predict the leakage failure forms and leakage amount, which meets the demand of multi-condition leakage prediction of water supply network.

关 键 词：漏失量预测 SAA-SVM模型 漏口失效形式 供水管网仿真模拟试验平台 

分 类 号：TV672.2[水利工程—水利水电工程] TU991[建筑科学—市政工程]