| 面向热轴故障的高速列车轴温阈值预测模型 |
| |
| 引用本文： | 谢国,王竹欣,黑新宏,高橋聖,望月宽.面向热轴故障的高速列车轴温阈值预测模型[J].交通运输工程学报,2018,18(3):129-137. |
| |
| 作者姓名： | 谢国 王竹欣 黑新宏 高橋聖 望月宽 |
| |
| 作者单位： | 1.西安理工大学 陕西省复杂系统控制与智能信息处理重点实验室, 陕西 西安 7100482.日本大学 , 千叶 船桥 274-8501 |
| |
| 基金项目： | 国家重点研发计划2017YFB1201500国家自然科学基金项目U1534208国家自然科学基金项目61773313陕西省重点研发计划2018GY-139 |
| |
| 摘 要： | 针对现有基于车轴温度固定阈值的故障检测系统适应性差且误报率、漏报率高的问题, 综合考虑列车速度、环境温度与运行工况等因素对轴温的影响以及各因素之间的关系, 建立了高速列车轴温动态阈值预测模型; 考虑高速列车在不同运行工况下轴温变化的差异特征, 将列车运行状态分为加速、匀速和减速3个阶段, 并针对每个阶段运用皮尔逊相关系数法分析列车速度、环境温度、荷载等原始监测数据以及各阶段运行时间、初始轴温等衍生数据与轴温的相关程度; 提取与轴温变化密切相关的因素, 基于多元回归分析方法, 针对列车的3个运行阶段, 分别建立基于原始监测数据的轴温动态阈值预测模型和基于原始监测数据与衍生数据的改进轴温动态阈值预测模型, 并采用F检验方法对模型的有效性进行检验, 基于中国高速列车实测轴温数据对模型的正确性进行了验证。研究结果表明: 列车在加速、匀速与减速3个阶段中, 轴温真实值与改进轴温动态阈值预测模型预测值的平均相对误差分别为2.0%、4.1%和3.3%;相对于基于原始监测数据的轴温动态阈值预测模型, 3个阶段中改进轴温动态阈值预测模型的预测精确度分别提高了79.8%、64.3%和65.6%;改进预测模型的决定系数大于0.99, 显著性概率小于0.05, 表明模型有效。
|
| 关 键 词： | 高速列车 热轴故障 数据驱动方法 轴温阈值预测 多元回归 |
| 收稿时间： | 2017-12-15 |
Axle temperature threshold prediction model of high-speed train for hot axle fault |
| |
| XIE Guo,WANG Zhu-xin,HEI Xin-hong,GAO Qiao-sheng,WANG Yue-kuan.Axle temperature threshold prediction model of high-speed train for hot axle fault[J].Journal of Traffic and Transportation Engineering,2018,18(3):129-137. |
| |
| Authors: | XIE Guo WANG Zhu-xin HEI Xin-hong GAO Qiao-sheng WANG Yue-kuan |
| |
| Affiliation: | 1.Shaanxi Key Laboratory of Complex System Control and Intelligent Information Processing, Xi'an University of Technology, Xi'an 710048, Shaanxi, China2.Department of Computer Engineering, Nihon University, Funabashi 274-8501, Chiba, Japan |
| |
| Abstract: | Aiming at the problem that the adaptability of existing fault detection system based on the fixed temperature threshold for the axle was poor, and its high false and missing alarm rate, considering the influence of train speed, environment temperature and running conditions on the axle temperature and the relationship among the factors, a dynamic threshold prediction model for the axle temperature of high-speed train was established. Considering the difference in axle temperature variation of high-speed train under different running conditions, the train running state was divided into three stages: acceleration, steady running and deceleration, and aiming at each stage, the Pearson correlation coefficient method was used to analyze the correlation degree between the axle temperature and original monitoring data of train speed, environment temperature and load, as well as that between the axle temperature and derivative data of runningtime and initial axle temperature. The factors closely related to axle temperature variation were extracted, the multiple regression analysis method was used to establish a dynamic threshold prediction model for axle temperature based on the original monitoring data, and a modified dynamic threshold prediction model based on the original monitoring data and derived data for the three running stages of the train. The models were validated using the Ftest method. The model accuracy was verified based on the measured axle temperature data from high-speed trains in China. Research result shows that in the three stages of acceleration, steady running and deceleration, the average relative errors between the true values of axle temperature and the prediction values of the modified dynamic threshold prediction model are 2.0%, 4.1% and 3.3%, respectively. The prediction accuracies of the modified prediction model in the three stage increase by 79.8%, 64.3%, and 65.6%, respectively, compared to the dynamic threshold prediction model for axle temperature based on the original monitoring data. The decision coefficient of the model is larger than 0.99 and the significance probability is less than 0.05, which indicates that the model is effective. |
| |
| Keywords: | |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《交通运输工程学报》浏览原始摘要信息 |
|
点击此处可从《交通运输工程学报》下载全文 |
|
