Enhancing the Performance of a Multiobjective Evolutionary Algorithm for Sanitary Sewer Overflow Reduction

摘要

The application of multiobjective evolutionary algorithms (MOEAs) to sanitary sewer overflow (SSO) optimization problems typically requires multiple runs of a simulation model and can be very computationally expensive. There is a need for simulation-optimization models that use fewer functional evaluations of the hydraulic model to identify near optimal solutions. In this study, two conflicting objectives were analyzed: maximizing SSO reduction and minimizing rehabilitation cost. This paper introduces a novel MOEA, the enhanced nondominated sorting evolution strategy (eNSES) that uses a specialized operator to guide the algorithm toward known SSOs locations. This strategy is being tested in an existing network in the eastern San Antonio Water System network. It has been compared with NSGA-II and NSES based on hypervolume and the overall nondominated vector generation ratio (ONVGR). The results show that eNSES improves the convergence rate by approximately 70% over the tested alternative algorithms, performing as well as NSGA-II and outperforming NSES in terms of the hypervolume by nearly 10%. In terms of the ONVGR, eNSES performed similarly to NSES but outperformed NSGA-II by 42%. (C) 2017 American Society of Civil Engineers.