Multi-Objective Cuckoo Search for the Optimal Design of Water Distribution Systems

Q. Wang,S. Liu,H. Wang,D. A. Savić
DOI: https://doi.org/10.1061/9780784412435.072
2012-01-01
Abstract:Multi-objective Evolutionary Algorithms (MOEAs), such as NSGA-II, has demonstrated encouraging strength on optimization during the past decade. Recently, cuckoo search, which was inspired by the obligate brood parasitism of some cuckoo species in nature, was proposed and reported to consistently outperform Genetic Algorithms and Particle Swarm Optimization on single-objective benchmark functions. In this paper, we combined cuckoo search with non-dominated sorting procedure of NSGA-II to generate a multi-objective cuckoo search (MOCS) algorithm. Then it was applied to the optimal design problems of Water Distribution Systems. The objectives took total cost and network resilience into account. The hydraulic simulations were implemented via EPANET software. To compare the power of MOCS and NSGA-II, the ratio of hypervolume (HVR) was used to assess the performance of each algorithm across multiple independent runs. The results obtained clearly show that MOCS is superior to NSGA-II by yielding more solutions of high quality in terms of convergence and diversity.
What problem does this paper attempt to address?