A vehicle model of eight degrees of freedom was established. Then, based on dynamic simulation model, an approximate model between design variables, uncertain variables and objective function was built. Using double-nested IP-GA, the ride comfort of the model was improved. For IP-GA, the interval of objective function was obtained through the inner layer while the optimal variable was searched through the outer layer. Finally, the results were compared with those of certain optimization, which has shown the superiority of uncertain optimization.