基于相关向量机的高速列车牵引系统剩余寿命预测

高速列车牵引系统在运行过程中总是受到诸多不确定因素的影响, 例如, 由于列车的负载、运行环境及元器件的老化引起的不确定性, 不确定因素不可避免地影响牵引系统剩余寿命的预测精度. 为了提高不确定情景下剩余寿命预测的准确性, 本文首先采用改进的相关向量机(Relevance vector machine, RVM)方法, 建立鲁棒性能良好的多步回归模型, 由于t分布比常用的高斯分布更具有鲁棒性, 通过权重和随机误差服从t分布而非高斯分布, 改进了相关向量机回归模型, 随后将超参数的先验一并融入似然函数, 通过最大化似然函数估计未知的超参数, 此外, 利用首达时间方法从概率角度对剩余寿命进行了预测, 最后通过牵引系统中电容器退化的案例, 与传统的相关向量机方法、自回归方法和支持向量机方法进行对比, 验证了所提算法的有效性.

Relevance Vector Machine Based Remaining Useful Life Prediction for Traction Systems of High-speed Trains

Traction systems often suffer from many uncertainties during their running processes, such as the inevitable uncertainties caused by the change of loading, operation and usage conditions. In order to improve the accuracy of remaining useful life (RUL) prediction under the uncertain scenario, a robust multi-step regression model is established by the improved relevance vector machine (RVM) method, in which weights and random errors are t distributed rather than Gaussian distributed. Then, unknown hyperparameters are estimated by taking priors of the hyperparameters into consideration. Moreover, the RUL is predicted by the first hitting time (FHT) method in probability perspective. The proposed method is demonstrated by a case study of capacitors degradation in traction systems. The results show the effectiveness of the proposed method.