Solution analysis in multi-objective optimization

Recent years have seen a growth in the use of evolutionary algorithms to optimize multi-objective building design problems. The aim is to find the Pareto optimal trade-off between conflicting design objectives such as capital cost and operational energy use. Analysis of the resulting set of solutions can be difficult, particularly where there are a large number (possibly hundreds) of design variables to consider. This paper reviews existing approaches to analysis of the Pareto front. It then introduces new approach to the analysis of the trade-off, based on a simple rank- ordering of the objectives, together with the correlation between objectives and problem variables. This allows analysis of the trade-off between the design objectives and variables. The approach is demonstrated for an example building, covering the different relationships that can exist between variables and the objectives.