Use of surrogate modelling for multiobjective optimisation of urban wastewater systems

摘要

Simulation models are now available to represent the sewer network, wastewater treatmentnplant and receiving water as an integrated system. These models can be combined withnoptimisation methods to improve overall system performance through optimal control.nEvolutionary algorithms (EAs) have been proven to be a powerful method in developing optimalncontrol strategies; however, the intensive computational requirement of these methods imposesna limit on their application. This paper explores the potential of surrogate modelling innmultiobjective optimisation of urban wastewater systems with a limited number of modelnsimulations. A surrogate based method, ParEGO, is combined with an integrated urbannwastewater model to solve real time control problems. This method is compared with thenpopular NSGA II, by using performance indicators: the hypervolume indicator, additive binaryn1-indicator and attainment surface. Comparative results show that ParEGO is an efficient andneffective method in deriving optimal control strategies for multiple objective control problemsnwith a small number of simulations. It is suggested that ParEGO can greatly improvencomputational efficiency in the multiobjective optimisation process, particularly for complexnurban wastewater systems.