A k, k-ε optimality selection based multi objective genetic algorithm with applications to vehicle engineering

Abstract In the paper, a multi objective genetic algorithm based on the concept of k-optimality and k-ε-optimality (KEMOGA) is introduced and applied. Pareto optimality alone is not always adequate for selecting a final solution because the Pareto optimal set can be very large. The k-optimality approach and the more general k-ε-optimality method, can be used to rank the Pareto-optimal solutions. The two methods have been included into a genetic algorithm selection procedure. The k-optimality method searches for points which remain Pareto-optimal when all of the subsets of n-k objectives (n is the number of objective functions) are optimised. The k-ε approach considers not only if an objective is worse than the others but also the entity of this variation through the introduction of a vector of indifference thresholds. The KEMOGA has been applied for the solution of two engineering problems. The selection of the stiffness and damping of a passively suspended vehicle in order to get the best compromise between discomfort, road holding and working space and a complex problem related to the optimisation of the tyre/suspension system of a sport car. The final design solution, found by means of the KEMOGA seems consistent with the solution selected by skilled suspensions specialists. The proposed approach has been tested and validated on a complex optimization problem. The solved problem deals with the optimization of the tyre/suspension system of a sport car. The proposed approach (KEMOGA) has shown to be very effective in terms of computational efficiency and accuracy.