Dear editor,rnWe propose a data-driven health monitoring method for running gears of a high-speed train.In the proposed approach, the principal component analysis (PCA) method is used to perform feature extraction while the belief-rule-base (BRB) method is used to perform health monitoring tasks.To improve the monitoring performance, constraints of the covariance matrix adaptive evolution strategy (CMA-ES) is used to optimize the monitoring parameters of the BRB method.Similarly, to verify the effectiveness of the proposed health monitoring approach, a set of real data of a high-speed train is used in the case studies.