关键词: 交通运输工程, 列车速度异常检测, 极端梯度提升, 交叉验证, 网格搜索, 格拉布斯检验

Abstract: With the rapid development of urban rail transit, it is necessary to study the detection method of abnormal train speed in order to ensure the safe running of trains. Therefore, a train speed anomaly detection method combining extreme gradient boost (XGboost) and anomaly detection method was proposed. Firstly, the XGboost model was trained by using the data of train speed sampled on site. Then, cross validation and grid search method were used to determine the optimal parameters of XGboost model. Finally, the maximum likelihood estimation and Grubbs test were used to determine the anomaly of the prediction results. The experimental results show that: compared with the other four common models in the test set, F1 values are increased by 7.08%, 12.9%, 16.9% and 2.9% respectively. Moreover, the proposed method meets the real-time detection requirements of train operation in terms of time efficiency.

Key words: traffic and transportation engineering, abnormal train speed detection, extreme gradient boosting, cross-validation, grid search, Grubbs test