Adaptive IDEA for Robust Multiobjective Optimization, Application to the r-TSALBP-m/A

Robust optimization tries to find flexible solutions when solving problems with uncertain scenarios and vague information. In this paper we present a multiobjective evolutionary algorithm (EMO) to solve robust multiobjective optimization problems. This algorithm is a novel adaptive method able to evolve separate populations of robust and non robust solutions during the search. It is based on the existing infeasibility driven evolutionary algorithm (IDEA) and uses an additional objective to evaluate the robustness of the solutions. The original and adaptive IDEAs are applied to solve the rTSALBP-m/A, an assembly line balancing model that considers a set of demand production plans and includes temporal overloads of the stations of the assembly line with respect to these plans as robustness functions. Our results show that the proposed adaptive IDEA gets more robust non-dominated solutions for the problem. Also, we show that, for the case of the r-TSALBP-m/A, we can obtain Pareto fronts with a higher convergence when including robustness information during the search of the algorithm.