城市配水管网区域改造聚类技术研究

Clustering-based Regional Transformation Technology for City Water Supply Network

阐述了模糊C均值(FCM)算法和马尔科夫随机场(MRF)的相关理论知识和基本框架。首先建立了基于模糊C均值的给水管网区域改造算法,它可直接应用MPR-Pipe算法得到的管网管段"重要性"值完成对管网的区域聚类;接着,建立了基于马尔科夫随机场和FCM算法的MRF-FCM区域改造算法,它同时考虑管网的拓扑信息和属性信息对管网区域聚类。根据工程实例进行算法实践,结果表明,两种算法可以得到符合实际工程应用的区域聚类,确定优先改造区域,为管网改造问题提供了切实可行的解决方案。

The relevant theoretical knowledge and basic framework of the fuzzy C-means algorithm （FCM algorithm） and Markov random field （MRF） were described. First, the regional transformation al- gorithm of water supply network based on fuzzy C-means was established, which could directly apply ＂im- portance＂ values from the MPR-Pipe algorithm to complete the regional clustering of pipe network. Then, the MRF-FCM regional transformation algorithm based on the Markov random field and FCM algorithm was established, which simultaneously considered the topological and attribute information of pipe network in regional clustering. The algorithms were tested according to the practical project, and the results showed that the two algorithms could figure out the regional clustering fit to actual project, and determine the priority for the transformation region to provide a practical solution for transformation issues of pipe network.