非均衡速度下的车轮踏面磨耗量预测研究  

Research on the prediction of wheel tread wear under unbalanced speed

作　　者：张国彪 王巍 Zhang Guobiao;Wang Wei(Baotou Vehicle Maintenance Branch,National Energy Railway Equipment Co.,Ltd.,Inner Mongolia Baotou,014040,China;Safety Supervision Department of China Energy Group,Beijing,100010,China)

机构地区：[1]国能铁路装备有限责任公司包头车辆维修分公司,内蒙古包头014040 [2]国家能源集团安全监察部,北京100010

出　　处：《机械设计与制造工程》2025年第2期134-138,共5页Machine Design and Manufacturing Engineering

基　　金：国能铁路装备有限责任公司科技项目(B31206220005)。

摘　　要：针对非均衡速度下车轮踏面磨耗量其预测准确度下降的问题,提出了非均衡速度下车轮踏面磨耗量预测方法。采用SMOTE技术对非均衡速度下车轮踏面磨耗数据进行聚类处理,将所得数据进行无量纲化处理后作为样本数据。构建线性回归磨耗量预测模型,引入XGBoost目标函数,累积训练样本集合,利用打分函数调节预测准确度。将样本数据输入预测模型中,得到车轮踏面磨耗量预测结果。在试验条件下,搭建试验平台并进行了非均衡速度下车轮踏面磨耗量预测试验。试验结果表明,采用该方法预测的车轮踏面平均磨耗率、磨耗量与实测结果有较高拟合度,验证了该方法的有效性。The variation of wheel tread wear at non-uniform speeds is unstable,leading to a decrease in prediction accuracy.Therefore,a method for predicting wheel tread wear at non-uniform speeds is proposed.Using SMOTE technology to cluster the wheel tread wear data at non-uniform speeds,the obtained data is dimensionless and used as sample data.It builds a linear regression wear prediction model,introduces XGBoost objective function,accumulates the training sample set,and uses scoring the function to adjust prediction accuracy.Based on inputting the sample data into the prediction model,it obtains the predicted result of wheel tread wear.Under the requirements of meeting the experimental conditions,a test platform is built to conduct predictive tests on wheel tread wear.The experimental results show that the predicted average wear rate and wear amount of the wheel tread by this method are basically consistent with the measured data,and the prediction accuracy of the wheel tread wear amount is high,proving that using this method for prediction results is more accurate.

关 键 词：非均衡速度 车轮踏面 磨耗量预测 SMOTE 过拟合 

分 类 号：TN92[电子电信—通信与信息系统]