供水管网水质监测传感器网络布局优化

摘要

To deal with the sudden pollutant injection,the optimization of layout of water quality monitoring sensors for water distribution network has been more and more concerned in research.To against the shortcoming of the layout optimization based on the concept of "service level" which is easily to be disturbed by subjective factors,the model of the shortest detection time for pollution incident is proposed.With the average detection time of the pollution incident as the optimization object,the manual setting of "service level" is unnecessary,the number of the detection points of the layout strategy provided by the model is fully determined by budget,once the number of monitoring points is given,the average detection time of the pollution incident is optimum,and the objective accuracy of the model can be guaranteed.The status of the water distribution network is averaging processed to improve the spatial utilization of calculation resources.In addition,because the commonly used optimizing algorithms of model parameters cannot satisfy the requirement for large scale water distribution network,the parallel PSO algorithm based on multi-threading mechanism is also proposed.This fully plays the capability of multi-core processor,and enhances the utilization of hardware resource and the model calculation speed.The test results indicate that the model and algorithm proposed possess certain reference significance for layout optimization of water monitoring sensors in water distribution network.%为应对突发污染物注入预防供水管网的问题,水质监测传感器网络布局优化已越来越多地受到研究关注.针对基于"服务水平"概念的布局优化模型易受主观因素干扰的问题,提出了污染事件检测时间最短模型.该模型以污染事件平均检测时间为优化目标,无需人为设定"服务水平",所得布局方案监测点数量完全由预算决定.一旦给定监测点数量,能够保证污染事件平均检测时间最优,充分保证了模型的客观准确性.对供水管网系统状态进行了平均化处理,提高了计算资源的空间利用率.同时,针对目前常用的模型参数寻优算法难以满足大规模供水管网模型求解的特点,提出了基于多线程机制的并行粒子群优化(PSO)算法,从而充分发挥多核心处理器的计算能力,提高了硬件资源的利用率,加快了模型计算的速度.试验结果表明,所提出的模型和优化算法对实际的供水管网水质监测网络布局优化选址具有一定的参考作用.