LEMMO: Hybridising Rule Induction and NSGAII for Multi-Objective Water Systems Design
Laëtitia Jourdan, David Corne, Dragan A. Savic, G. A. Walters · 2005
Many studies use genetic algorithms to model water distribution network and they give some interesting results. Recent studies propose a multi-objective model of the problem. But one of the drawbacks of genetic algorithm both mono and multi-objective is the extensive use of the evaluation process. In water systems design, the evaluation of the quality of a water distribution network requires a time expensive simulation. These article deals with the use of machine learning to boost the convergence of a multi-objective genetic algorithm in the particular context of water systems design. By testing the method called LEMMO, on small and large networks we prove the interest of our algorithm to accelerate the search of a multi-objective genetic algorithm.