基于可靠性的给水管网传感器多目标选址

针对蓄意污染事件下给水管网传感器优化选址问题,提出了以污染事件监测比例最大化、污染事件监测时间期望值最小化和选址方案可靠性最大化为设计目标的多目标优化选址模型. 结合一管网算例,利用EPANET2模拟污染入侵后管网结点的污染物浓度时间序列,通过蒙特卡洛方法获取污染事件监测时间矩阵；应用基于整数编码的非支配排序遗传算法-II (NSGA-II)求解该模型,给出了设计目标间的Pareto前沿面并对参数敏感性做了分析. 结果表明,传感器预算数目与可靠性时间标准对传感器选址结果有着显著影响. 提出的模型可用于对给水管网传感器选址问题的多个竞争目标进行权衡分析．

Reliability-based multi-objective sensor location in water distribution systems

A multi-objective optimal location model was put forward for sensor placement in water distribution systems under deliberate contamination events. Three design objectives were taken into account, including maximizing the detection likelihood, minimizing the expected detection time and maximizing the reliability of location plan. The contaminant concentration series at each node were simulated using EPANET2 when the water distribution system was suffered from contaminant injection, then the detection time matrix was generated based on Monte Carlo method. An improved non-dominated sorted genetic algorithm-II (NSGA-II) with integer code was proposed to solve this model. Pareto fronts were plotted among design objectives and sensitivity analysis was implemented. Both the number of sensors and the reliability criterion are significant to sensor location. The proposed location model  is constructive for exploring trade-offs among multiple conflicting objectives for sensor location  in water distribution systems．