The invention provides a train traction energy consumption anomaly detection method based on Boosting and multi-predictor fusion. The method comprises the following steps: processing second-level accumulated train traction energy consumption data to form a time sequence data set of day, week, month and year time scales; a plurality of time sequence predictors are constructed, each time sequence predictor is verified, a Boosting algorithm is adopted to adjust the prediction result weight of each time sequence predictor, and a multi-predictor fusion model is formed; inputting second-level accumulated train traction energy consumption data of a train to be analyzed into the multi-predictor fusion model, and judging the traction energy consumption state of the train to be analyzed according to an output value of the multi-predictor fusion model. The anomaly judgment basis provided by the invention depends on a plurality of time sequence prediction models and Boosting algorithm optimization weight, and the obtained predicted value is weighted and verified, conforms to the actual energy consumption level of an engineering site, and is convenient for workers to understand and use at the same time.
本发明提供了一种基于Boosting和多预测器融合的列车牵引能耗异常检测方法。该方法包括:处理秒级累计的列车牵引能耗数据,形成日、周、月和年时间尺度的时序数据集;构建多种时间序列预测器,对每种时间序列预测器进行校验验证,采用Boosting算法调整各时间序列预测器的预测结果权重,形成多预测器融合模型;将待分析列车的秒级累计的列车牵引能耗数据输入到所述多预测器融合模型,根据所述多预测器融合模型的输出值判断所述待分析列车的牵引能耗状态。本发明提出的异常判断基础依托于多种时间序列预测模型及Boosting算法优化权重,所得预测值经加权及校验,符合工程现场实际能耗水平,同时便于工作人员理解使用。
Train traction energy consumption anomaly detection method based on Boosting and multi-predictor fusion
基于Boosting和多预测器融合的列车牵引能耗异常检测方法
25.11.2022
Patent
Elektronische Ressource
Chinesisch
Determination of Energy Consumption for Train Traction
Springer Verlag | 2022
|Europäisches Patentamt | 2025
|Train energy consumption control method based on multi-train cooperation
Europäisches Patentamt | 2022
|