Abstract:In the process of high-intensity transportation, heavy-haul railways are prone to damage such as stripping and abrasion of rails, which affects driving safety. In order to ensure the safe operation of the railway, it is very important to monitor the damage status of the rail. However, at present, the rail damage detection method is mainly based on manual road inspection, and the detection results have problems such as strong subjectivity, difficulty in quantifying the degree of damage, and difficulty in predicting the trend of damage degradation. In order to solve the existing problems, this paper proposes a new method based on PCA and TCN-Attention to predict the degradation trend of stripping damage of heavy-haul railway rails. Firstly, the time-domain and frequency-domain features were extracted from the rail stripping damage vibration signal, and the high-dimensional features were reduced by PCA. Secondly, by using the difference of characteristics between time series samples, the degradation index of rail stripping damage was constructed to describe the degradation trend, which solved the problem of difficulty in measuring the damage state. The TCN network model combined with the attention mechanism was used to pay attention to the effective features to improve the prediction accuracy of the model. Finally, the effectiveness of the proposed method is verified by using the vibration data of the whole life cycle of the rail collected by a railway locomotive depot from normal to damage to failure, and the experimental results show that the proposed method can accurately predict the degradation trend of rail stripping damage.