A time series prediction method based on the dynamic Bayesian least squares support vector machine(LS-SVM) is proposed to accurately predict operating parameters of the turbofan engine. By the Bayesian evidence framework theory, initial model parameters of the LS-SVM are inferred. Dynamic learning of the LS-SVM and dynamic prediction of time series are realized by the recursively incremental and decremental sample learning method. The method is applied to dynamically predict frictional moment time series of a certain type of the turbofan engine, and a comparison is made with the dynamic LS-SVM. The results show that the proposed method is effective in improving prediction accuracy.