基于VMD-ICSO-GRU的高铁列控车载设备故障率时间序列预测

Time Series Prediction of Fault Rate of High-speed Railway On-board Equipment Based on VMD-ICSO-GRU

有效地预测高铁列控车载设备故障率对合理分配设备备品、制定维修计划、减少故障发生具有重要意义。以列车运行控制系统的历史故障数据为对象,提出一种基于变分模态分解(VMD)和门控循环单元(GRU)的故障率预测模型。首先,利用VMD将车载设备故障率时间序列分解为一组包含不同频率信息的子序列,降低原始序列的非平稳性;然后,针对分解后的各个子序列建立多个基于GRU的时间序列预测模型,为提高预测精度,提出一种改进的猫群优化(ICSO)算法自适应设置各个GRU网络参数;最后,叠加各子序列预测结果得到最终故障率预测值。收集CTCS3-300T型列控车载设备历史故障数据进行实验,结果表明,相比于其他时间序列预测模型,本文模型得到的均方根误差(RMSE)和平均绝对误差(MAE)分别为0.0445和0.0391,均低于其他模型,验证了其有效性。

Effective prediction of the failure rate of high-speed railway on-board equipment is of great significance for rational allocation of equipment spare parts,formulation of maintenance plan and reduction of occurrence of failures.According to the historical fault data of train operation,a fault rate prediction model based on variational mode decomposition(VMD)and gate recurrent unit(GRU)was proposed.Firstly,the time series of on-board equipment failure rate was decomposed into a group of subsequences containing different frequency information by using VMD,so as to reduce the non-stationarity of the original sequence.Then,several GRU-based time series prediction models were established for each decomposed subsequence.In order to improve the prediction accuracy,an improved cat swarm optimization(ICSO)algorithm was proposed to adaptively set the parameters of each GRU.Finally,the final failure rate prediction value was obtained by superimposing the prediction results of each subsequence.The historical fault data of 300T on-board equipment were collected for experiment,and the experimental results show that compared with other time series prediction models,the root mean square error(RMSE)and mean absolute error(MAE)obtained by the proposed model are 0.0445 and 0.0391 respectively,which are lower than other models,indicating the effectiveness of the proposed model.