结合模型校核和贝叶斯决策理论的管网漏失定位  

Leakage Location of Water Distribution System Based on Model Calibration and Bayesian Decision Theory

作　　者：谭奇峰 高金良[3] 叶健[3] 乔仪超 涂敏[3] 

机构地区：[1]广东粤海控股集团有限公司,广东深圳518021 [2]广东粤港供水有限公司,广东深圳518021 [3]哈尔滨工业大学市政环境工程学院,黑龙江哈尔滨150090

出　　处：《中国给水排水》2016年第19期44-48,共5页China Water & Wastewater

基　　金：广东省教育部产学研结合项目(2011A090200040);国家自然科学基金资助项目(51278148);欧盟第七框架"玛丽.居里行动"计划项目(PIRSES-GA-2012-318985)

摘　　要：管网中发生漏失事故时必然会引起其他节点的水压波动,波动程度随漏失位置和漏失流量而变化。基于这一思想,采用模型校核和贝叶斯决策理论分别建立漏失定位模型,分析了两个模型的原理及优缺点,并在此基础上提出一种改进漏失定位模型,该模型对水力模型误差和测量误差有一定的免疫力,而且具有时效性。以CP市供水管网为例,通过消火栓放水模拟漏失,求解改进的漏失定位模型得到漏失概率,从而能确定出漏失节点所在区域,达到较高的检测精度。Pressure fluctuations of nodes can be caused by leakage accident in pipe network. Extent of fluctuations changes with the changes of leakage location and leakage flow. Based on this, the model calibration and Bayesian decision theory were adopted to establish leakage location models respectively. The principles, advantages and disadvantages of the two models were analyzed, and an improved leakage location model was proposed. The improved model has some immunity for the model error and measurement error, and it has timeliness. This model was illustrated by the example of water distribution system in Changping Town, Dongguan City, Guangdong Province. The burst accident was simulated by draining of fire hydrant, and then the improved model was solved to get probability of burst, namely the location and the flow of the burst. It has very high detection accuracy.

关 键 词：城镇供水管网 漏失定位 模型校核 贝叶斯决策理论 遗传算法 

分 类 号：TU991.33[建筑科学—市政工程]