基于信息最大化准则的供水管网压力监测点布置

系统监测的目的是为了诊断或预报系统状态. 为保证监测点提供的是系统的有效监测信息，而不是冗余信息, 从而提高诊断预报的准确性, 监测点不仅应具有灵敏性, 且监测点间的信息相关性越小越好. 为了实现监测信息最大化, 在供水管网灵敏度分析的基础上, 分别采用有效独立法、Fisher信息矩阵最大化准则, 以及节点相关系数3种方法确定管网监测点. 最后, 通过对某一供水管网的压力监测点布置, 将采用3种方法和传统聚类分析方法得到的方案进行了对比分析. 分析表明: 基于节点相关系数的监测点布置方法能同时得到监测点和监测域, 且算法简单、稳定, 优于其他方法.

Layout of monitoring points in water supply network based on information maximization criterion

The objective of system monitoring is mainly for diagnoses or forecasting. Monitoring points should be highly sensitive, and their outputs should have low correlation to avoid information redundant, therefore the obtained information is more useful in reliable diagnoses or forecasts. To maximize usefulness of the monitoring information, an effective independence method, a maximization criterion of Fisher information matrix, and correlation coefficients are used to determine the monitoring points based on the sensitivity analysis of water supply network. Finally, a water supply network is analyzed, and schemes of the three methods compared with that of the traditional cluster analysis method. It is shown that the monitoring points and their corresponding monitoring areas can be determined using the method of monitoring point locating based on correlation coefficients. The method is more simple, stable and efficient, so it is better than other methods.