基于LSTM-BPNN-SVR的地铁车辆轴箱温度预测方法  

Prediction Method of Railway Vehicle Axle Box Temperature Based on LSTM-BPNN-SVR Algorithm

作　　者：张玮东 ZHANG Weidong(Vehicle Branch,Shanghai Metro Maintenance Guarantee Co.,Ltd.,Shanghai 200031)

机构地区：[1]上海地铁维护保障有限公司车辆分公司,上海200031

出　　处：《计算机与数字工程》2022年第9期2079-2082,2108,共5页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:51907117,51975347);上海市科技计划项目(编号:22010501600);上海申通地铁集团重点项目(编号:KY20R013-3,JS-KY21R008-6)资助。

摘　　要：针对上海地铁所使用的自发电式列车轴箱温度感知及预警系统中样本数据非等距及高精度预测问题,通过对采样数据特征分析,利用K均值聚类及分段三次Hermite插值法,实现了样本数据的等距化;搭建LSTM-BPNN-SVR预测模型强化了对样本数据特征信息的提取,实现了对地铁车辆轴箱温度高精度预测的目的。通过与实测数据进行对比,该模型预测精度可达到99.15%,验证了该方法的有效性;通过与其他六种常规模型的预测结果对比,LSTM-BPNN-SVR预测模型具有更高的预测性能。Aiming at the problems of non-equidistant sample data and high-precision prediction of monitoring sensor data in the self-generating rail vehicle axle box temperature monitoring and warning system used by Shanghai Metro,the equidistance of sample data is realized by analyzing the characteristics of sampling data,using k-means clustering and piecewise cubic Hermite interpolation method. By building the LSTM-BPNN-SVR prediction model,the extraction of the feature information of the sample data is strengthened,and the purpose of high-precision prediction of the temperature of the axle box of the subway vehicle is realized.Compared with the measured data,the prediction accuracy of the model can reach 99.15%,which verifies the effectiveness of the method. Compared with the prediction results of other six conventional models,LSTM-BPNN-SVR prediction model has higher prediction performance.

关 键 词：列车轴温 非线性采样 数据融合 组合预测模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]