Optimal operation of pumping stations in Water Distribution Networks (WDNs) is presented based on three conflicting objectives, including 1) pumping energy costs, 2) hydraulic reliability, and 3) quality reliability. We proposed a three-objective optimization algorithm called Clustered Non-dominated Archiving Ant Colony Optimization (Clustered-NA-ACO) and evaluated the efficiency of this algorithm through DTLZ test functions. Incorporating k-means clustering strategy with NA-ACO can greatly improve the distribution of solutions and reduce the run time compared to NA-ACO. A pressure-driven analysis tool called EPANET-Iterative Modifications to Nodal Outflows (EPANET-IMNO) is developed for dynamic analysis of WDNs and linked directly to Clustered-NA-ACO. We applied this simulation-optimization tool to two WDNs. The results show that it is necessary to consider hydraulic reliability and quality reliability as two separate objective functions in WDNs with storage tanks. We also suggested the combined form of the two reliability indicators in the optimization of WDNs without any storage tanks.

基于三个相互矛盾的目标，提出了水分配网络（WDN）中泵站的最佳运行，包括1）泵的能源成本，2）水力可靠性和3）质量可靠性。我们提出了一种三目标优化算法，称为聚类非支配档案蚁群优化（Clustered-NA-ACO），并通过DTLZ测试功能评估了该算法的效率。与NA-ACO相比，将k-means聚类策略与NA-ACO结合可以大大改善解决方案的分布并减少运行时间。开发了一种压力驱动的分析工具，称为EPANET节点流迭代修改（EPANET-IMNO），用于WDN的动态分析，并直接与Clustered-NA-ACO链接。我们将此仿真优化工具应用于两个WDN。结果表明，在带有储罐的WDN中，必须将水力可靠性和质量可靠性视为两个独立的目标函数。我们还建议在不使用任何储罐的情况下优化WDN时将两个可靠性指标的组合形式。