Paper Conference

Proceedings of Building Simulation 2011: 12th Conference of IBPSA


Optimisation of low-energy building design using surrogate models

Esmond Gerald Tresidder, Yi Zhang, Alexander Forrester

Abstract: This paper sets out to test the performance of surrogate modelling optimisation techniques on a highly-discrete building design problem. The true optimum is first established through comprehensive analysis using a multi-processor computer, this enables the performance of the optimisation to be assessed precisely. The performance is compared against a stand-alone evolutionary algorithm (EA) on the same design problem. Preliminary results show that for this design problem surrogate modelling is capable of finding the true optimum, and suggest that it may be more reliable and faster at doing so than a stand-alone EA.
Pages: 1012 - 1016