摘要
机车车轮轮缘厚度是评价机车运行安全的重要指标,为满足车轮运维单位对轮缘厚度的监测需求,提出基于LASSO-DBSCAN的机车车轮轮缘厚度尺寸动态测量数据处理算法。首先,基于人工测量数据,采用LASSO回归算法,并结合GOA优化算法进行参数调优,获取车轮轮缘厚度的变化趋势。随后,利用DBSCAN聚类算法标记并修正离群数据,建立滑动窗口降低数据波动。最后,基于最近一次的轮缘厚度人工测量数据,采用数量为10片车轮的样本数据,测试尺寸动态测量数据处理算法的效果。测试结果表明:车轮轮缘厚度尺寸动态测量的原数据绝对误差均值为0.32 mm,标准差为0.13 mm;经算法处理后,绝对误差均值下降至0.12 mm,标准差降至0.06 mm。经算法处理后的尺寸动态测量数据具有更高的精度和更好的稳定性。
The locomotive wheel rim thickness is an important indicator for evaluating the safety of locomotive operation.In order to meet the monitoring needs of wheel maintenance units for rim thickness,a locomotive wheel rim thickness dimension dynamic measurement data processing algorithm based on LASSO-DBSCAN was proposed.Firstly,based on manual measurement data,the LASSO regression algorithm was used,and combined with GOA optimization algorithm for parameter optimization to obtain the trend of wheel rim thickness changes.Subsequently,the DBSCAN clustering algorithm was used to label and correct outlier data,and a sliding window was established to reduce data fluctuations.Finally,based on the most recent manual measurement data of wheel rim thickness,using a sample of 10 wheel pieces,the effectiveness of the dynamic measurement data processing algorithm was tested.The test results show that the original absolute error mean of the dynamic measurement of wheel rim thickness was 0.32 mm,with a standard deviation of 0.13 mm;after processing by the algorithm,the absolute error mean decreased to 0.12 mm,and the standard deviation decreased to 0.06 mm.The dimension dynamic measurement data processed by the algorithm exhibits higher accuracy and better stability.
作者
孙宇铎
程亚萍
刘通
冯颖
王菲儿
王峰
SUN Yuduo;CHENG Yaping;LIU Tong;FENG Ying;WANG Feier;WANG Feng(Metals and Chemistry Research Institute,China Academy of Railway Sciences Corporation Limited,Beijing 100081,China;Beijing Advanced Materials Technology Co.,Ltd.,Beijing 100081,China)
出处
《高速铁路新材料》
2024年第5期44-47,共4页
Advanced Materials of High Speed Railway
基金
中国国家铁路集团有限公司科技研究开发计划(P2023J007)
中国铁道科学研究院集团有限公司科研项目(2022YJ245)。
