Optimization And Simulation Of Vehicle Suspension Parameters Base On PSO

Optimization of suspension parameters is an important research in the chassis design. It aims to improve the comfort, handling stability and security of automobile. It belongs to a typical multi-objective optimization problem, whose objectives are in conflict with each other. Among the different multi-objective optimization algorithms, many studies shows that the evolutionary algorithms includes particle swarm optimization and genetic algorithms is the most effective way to solve multi-objective problem. The PSO algorithm which has no genetic coding, crossover, compiler and other complex operations has a better convergence speed. In order to solve the problem of suspension Parameters'Optimization, the PSO algorithm was improved, and then the results were obtained. Finally, the suspension's Parameters, which obtained by the new algorithm and verified by the experiments, greatly improved the comfort of automobile.The main contributions of this dissertation can be summarized as follows:Firstly, I created the vehicle's motion equations with seven DOFs based on automotive vibration theory and established its MATLAB model. I choose suspension stiffness and damping as design variables.The objective function were determined as the vertical acceleration of body, dynamic travels and wheel-road dynamic displacement etc.Secondly, to make better use of particle swarm algorithm to achieve multi-objective optimization, based on the current study, this project proposed multi-objective decision based on the Vague cross entropy that got the value of pre-decision that suspension optimized the objective function by integrating the conflicting evaluation results of experts and solved the the excessive loss of convergence caused by losing diversity of particles, by applying the value of pre-decision to information Communications of the assistant swarm.Moreover, an inertia weight factor was proposed for cyclical attenuation adaptive strategies, to solving the balance between global and local search ability in particle swarm algorithm. Then, instead of the traditional criteria for classification of non-dominated Pareto optimal solutions, the effective-order search strategies were utilized. A better "quality" solution was selected from non-dominated solution efficiently, and the problem in the iterative algorithm of global optimal "quality" descent was solved. The standard library functions of the algorithm performance metrics were adopted to verify the validity and flexibility of strategies.Finally, I adopte the pre-decision MOPSO algorithm to optimize the suspension parameters. The set of solutions performances better than which is with the initial parameters in experiment. The results show that the ride comfort is significantly improved. Then I use objective random weighted sum method and genetic algorithm to solve the suspension problem, and do some comparison with the aforementioned results. It does prove to be much more effective than the traditional algorithm compared with the pre-decision MOPSO optimal solutions...