Genetic Algorithms for Optimal Design of Vehicle Suspensions

Abstract

The optimal design of vehicle suspensions is a complex design optimization problem which may have highly nonlinear design spaces with many local optima. At present, there are two main problems for the optimal design of vehicle suspensions. The first is that the global optimization solution is hardly reached using the numerical methods or the simple genetic algorithms. The second is that the suspension models are all lumped mass models in genetic algorithms. Virtual prototyping software can deal with complex multibody models and provide a good approximation to real system performance in different simulated scenarios, but the optimization method is a numerical method, thus the complex model gets local optimal solutions. So an ideal approach is to use virtual prototyping as the individuals performance evaluation means and to optimize the design based on GAs (genetic algorithms). The paper completes the interface between GAs and the ADAMS (Automatic Dynamic Analysis of Mechanical Systems) software, and presents an improved genetic algorithm for the optimal design of vehicle suspensions. In this method, a new scheme of genetic algorithms is used first, thus no crossover and mutation rates are needed. Father-offspring combined elitist scheme and generation replacement strategy are introduced to ensure a stable convergence of the algorithm. The optimization results prove that the improved genetic algorithm is better than the numerical optimization method, the simple genetic algorithm and the niching genetic algorithm in the aspects of gaining global optimization solutions and accelerating convergence