Uncertainty based evolutionary optimisation

This paper presents a robust evolutionary optimisation approach for real life design problems characterised by uncertainty. The proposed approach handles uncertainty in the design space, as well as in the objective functions and constrains, thanks to a new Pareto dominance criterion based on the neighbourhood around a solution. The approach is applied on a gearbox design optimisation problem as a case study. A comparison between two approaches, robust Pareto dominance criterion and a preference based penalty function, for deal with noisy environment is done for highlight the strength of the proposed robust Pareto dominance criterion.