The gearbox is one of the key components of the high-speed train system. In order to predict the fatigue damage behavior of high-speed train gearbox shell material, we propose a new method. Because the fatigue process is long and there is no whole life fatigue damage data of the shell, the performance degradation method and acceleration test with acoustic emission instrument have been used. A new Adaboost data distribution adjustment algorithm is proposed to solve the problem of data imbalance of acoustic emission signals during the fatigue process. Then a cumulative count trend model and a fatigue life prediction model are developed. The life prediction error is controlled within 400 s, most errors are less than 200 s, and the relative error is less than 1.1%, which ensures the feasibility of the developed model. In the future, this model can be used in long-time life prediction researches.

变速箱是高速列车系统的关键组件之一。为了预测高速列车变速箱壳体材料的疲劳损伤行为，我们提出了一种新的方法。由于疲劳过程很长，并且没有壳体的整个寿命的疲劳损伤数据，因此已使用性能下降方法和使用声发射仪进行的加速试验。针对疲劳过程中声发射信号的数据不平衡问题，提出了一种新的Adaboost数据分布调整算法。然后建立了累积计数趋势模型和疲劳寿命预测模型。寿命预测误差控制在400 s以内，大多数误差小于200 s，相对误差小于1.1％，这保证了所开发模型的可行性。将来，