In this study, we propose the ensembles of surrogates for an industrial application of fatigue optimization problem that aims to maximize a truck cab's fatigue life. After validating the numerical model, different ensembles of surrogates comprised polynomial response surface (PRS), radial basis function (RBF) and Kriging (KRG) models are established to approximate the fatigue life function. A hybrid PSO algorithm, which integrates the standard PSO with sequential quadratic programming (SQP), is implemented here to seek a quasiglobal optimum. Compared with individual surrogates, the ensembles of surrogates can attain more competent optima and yield a smaller surrogate error at the optimal point. Moreover, the hybrid PSO technique proves to search the better optima than the standard PSO in the fatigue optimization problem considered. Finally, it is found that a more accurate surrogate model may not necessarily produce a better optimum for the ensembles of surrogates, thus multiple ensembles are recommended without increasing much extra computational cost.