DART algorithm for constrained optimisation with applications in suspension design

Abstract

We propose a new hybrid algorithm with application of suspension designs problem. The algorithm is based on divide-and-rule technique (DART) governed by artificial bee colony (ABC) algorithm. A divide phase splits the search region into subregions of smaller size and a ruling phase allows bees to rule over the selected subregion and look for a new region with best food source. The DART process is recursively applied to discover higher fitness subregion producing the optimal solution to the problem. Moreover, a new search strategy has been implemented to bias the bees towards feasible solution. A suspension problem has been formulated as a nonlinear constrained optimisation problem where the weighted sum of sprung mass jerks and their deflections are minimised under technological constraints. By the proposed algorithm the optimal passive suspension parameters have been obtained from the half car model to minimise the level of vibration and enhance passengers comfort. Moreover, the results obtained by model simulations with existing and optimised suspension parameters are compared.