摘要
为准确预测高速列车的D电缆剩余使用寿命,保证列车安全、高效运行,提出1种基于SA-LSTM的高速列车D电缆剩余使用寿命预测方法。首先,分析D电缆的内部结构、主要失效模式和失效机理,构建D电缆的有限元模型;其次,基于热击穿失效机理,构建D电缆在不同热应力条件下的全生命周期加速退化数据集;然后,通过随机森林(RF)算法主动筛选关键退化特征,引入自注意力(SA)机制,融合统计特征并且结合长短时记忆(LSTM)网络实现剩余使用寿命(RUL)的准确预测;最后,通过D电缆退化模拟数据集验证所提方法的有效性。结果表明:所提方法性能表现优异,相较于传统LSTM,GRU,SVR和FNN等方法,在6种典型热应力工况下的平均绝对误差降低52.4%、均方根误差平均值降低49.7%,提高了复杂工况下D电缆的RUL预测精度。该方法对于提高铁路风险防控水平、保障列车运行安全具有参考价值。
In order to accurately predict the remaining useful life of D-cable for high-speed train and ensure the safe and efficient operation of trains,a prediction method of remaining useful life of D-cable for high-speed train based on SA-LSTM is proposed.Firstly,the internal structure,main failure mode and failure mechanism of Dcable are analyzed,and the finite element model of D-cable is constructed.Secondly,based on the thermal breakdown failure mechanism,the accelerated degradation data set of whole life cycle for D-cable under different thermal stress conditions is constructed.Then,the random forest(RF)algorithm is used to actively screen the key degradation features,and the self-attention(SA)mechanism is introduced to fuse statistical features and combine the long short-term memory(LSTM)network to achieve accurate prediction of remaining useful life(RUL).Finally,the effectiveness of the proposed method is verified by the D-cable degradation simulation data set.The results show that compared with the traditional methods such as LSTM,GRU,SVR and FNN,the proposed method has excellent performance.Under 6 typical thermal stress conditions,the average absolute error is reduced by 52.4%and the average root mean square error is reduced by 49.7%,which improves the RUL prediction accuracy of D-cable under complex working conditions.This method has reference value for improving the level of railway risk prevention and control,and ensuring train safe operation.
作者
柴琳果
张平
张辉
上官伟
陈俊杰
王剑
蔡伯根
CHAI Linguo;ZHANG Ping;ZHANG Hui;SHANGGUAN Wei;CHEN Junjie;WANG Jian;CAI Baigen(School of Automation and Intelligence,Beijing Jiaotong University,Beijing 100044,China;State Key Laboratory of Advanced Rail Autonomous Operation,Beijing 100044,China;Beijing Engineering Research Center of EMC and GNSS Technology for Rail Transportation,Beijing 100044,China;Signal and Communication Research Institute,China Academy of Railway Sciences Corporation Limited,Beijing 100081,China)
出处
《中国铁道科学》
北大核心
2026年第1期171-184,共14页
China Railway Science
基金
中央高校基本科研业务费专项资金资助项目(2025JBZX039)
北京市自然科学基金资助项目(L231003)
北京高校卓越青年科学家计划项目(JWZQ20240101010)
中国国家铁路集团有限公司科技研究开发计划课题(K2023S020)。
关键词
剩余使用寿命预测
D电缆
疲劳裂纹扩展
有限元分析
自注意力机制
Remaining useful life prediction
D-cable
Fatigue crack expansion
Finite element analysis
Selfattention mechanism
作者简介
第一作者:柴琳果(1988-),男,湖北荆门人,副教授。E-mail:lgchai@bjtu.edu.cn;通讯作者:上官伟(1979-),男,陕西咸阳人,教授。E-mail:wshg@bjtu.edu.cn。
