An assessment of a genetic algorithm-based approach for optimising multi-body systems with applications to vehicle handling performance

摘要

Simulation is seen as a way to develop an excellent understanding of vehicle dynamic behaviour and the utilisation of this understanding in designing vehicles. An essential component of this approach is the use of mathematical optimisation methods as a means of synthesising complex mechanical systems such as vehicles. Complex vehicle multi-body models have a significant number of design variables. Vehicle performance is assessed using a number of different performance measures calculated over a number of different scenarios. The resulting problem is one of significant complexity and deterministic optimisation methods are not suitable for this type of problem because of their propensity to lead to local minima. Stochastic Optimisation methods are not as sensitive to local minima and frequently yield better results, This paper presents the use of a Genetic Algorithm-based (GA) approach to vehicle handling design problems and compares the results obtained with other methods such as Simulated Annealing and Monte-Carlo. The results show that the Genetic-Algorithm approach shows distinct advantages over the other two methods. Results obtained using GAs show superior improvements in the vehicle handling performance over three manoeuvres.