摘要
基于北方某城市供水管网GIS数据库,首先对管网爆管基础数据进行分析与分类,并采用Logistic广义线性模型对管网爆管危险率进行预测研究;然后应用建立的供水管网Logistic爆管预测模型进行预测,并结合GIS系统的空间分析功能对模型进行检验。结果表明,当模型判别概率设为0.75时,其预测精度>90%,说明该模型具有较好的拟合效果和较高的预测精度,能够有效地筛选出爆管危险率较大的管道。
Based on the GIS database for the water distribution system in a northern city, the pa- rameters of pipe breaks were analyzed and classified. The breakage hazard rates were predicted by a Lo- gistic regression model (Logistic GLM). The Logistic prediction model for pipe breaks was established and evaluated with the space analysis functionalities of GIS. The results showed that the forecasting accu- racy was above 90% when the determinant probability of the proposed model was set to 0.75. The model demonstrated acceptable fitting rates and prediction accuracy and was effective for managing those pipes with high breaking hazard rates.
出处
《中国给水排水》
CAS
CSCD
北大核心
2012年第23期68-71,共4页
China Water & Wastewater
基金
国家自然科学基金资助项目(50878140
50778121)
关键词
供水管网
爆管
预测模型
LOGISTIC回归
water distribution system
pipe break
prediction model
Logistic regression