基于NSGA-Ⅱ的给水管网传感器多目标优化选址研究

Multi-objective Optimal Placement of Sensor for Water Distribution Systems based on NSGA-Ⅱ

在EPANET2污染物侵入模拟的基础上，以污染事件期望监测时间最短和传感器系统能够监测到的污染事件比例最大为优化目标，研究了传感器选址的多目标优化问题。针对选址问题的特性。通过改进非支配排序遗传算法-Ⅱ（NSGA-Ⅱ）的编码方式、交叉算子和变异算子，使其能够较好地解决给水管网传感器多目标优化选址问题。结合一管网算例，利用改进过的NSGA-Ⅱ给出了一组Pareto解．并对两个极端解对应的选址结果进行了权衡分析。提出的优化模型与求解方法对给水管网传感器多目标选址具有一定的参考作用。

The deliberate contamination intrusion to water distribution systems was simulated using EPANET2,then multiobjective optimal placement of sensor was studied focused on two objectives,one was minimizing the mathematical expectation of detection time for all contamination events and the other one was maximizing the detection likelihood of contamination events. Because of the special characteristics of sensor placement problem,this paper modified the code, crossover and mutation operators of NSGA-Ⅱ in order to resolve this problem better. With the help of a case,the Pareto solutions were generated using modified NSGA-Ⅱ and the sensor placements of extreme two solutions were analyzed. The optimal model and modified NSGA-Ⅱ proposed in this paper are conductive to multi-objective optimal placement of sensor for water distribution systems.