基于集对分析的供水管网漏损预测模型

Leakage forecast model of water supply network based on set pair analysis

探讨了集对分析法在供水管网漏损预测中的实现途径,从同、异、反三个方面度量漏损预测样本与历史样本的相似性,建立了基于集对分析的漏损预测(SPA-LF)模型。应用结果表明,SPA-LF对预测过程不确定性的描述较为详细,供水管网漏点数及漏损频率预测结果均显示出SPA-LF模型的有效性,减小了预测过程的不确定性,与灰色系统模型、指数平滑模型及神经网络模型预测结果相比,SPA-LF模型取得了满意的预测精度。SPA-LF模型计算简单,使用方便,可为供水管网的维护及管道更新提供决策支持。

The set pair analysis method was discussed for leakage forecast of water supply network. The similarity between the leakage prediction samples and the historical samples was measured from three aspects of identical, discrepancy and contrary, and the set pair analysis for leakage forecast （SPA-LF） model was established. The application results showed that the characterization of uncertainty of SPA-LF was detailed. The quantity of leakages in water supply network and the forecast results of leakage frequency showed that SPA-LF was effective, the uncertainty could be reduced in the course of forecast. A higher forecast precision was obtained with SPA-LF compared with grey system model, exponential smoothing model and neural network model. The calculation of SPA-LF was simple. Its application was convenient and it could provide decision support for water supply network maintenance and pipelines update.