Remaining useful life estimation of metropolitan train wheels considering measurement error

Jiang, Zhenan; Banjević, Dragan; Mingcheng, E; Jardine, Andrew; Li, Qi

doi:10.1108/jqme-04-2016-0017

Cited by 7 publications

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover,ẑ t is the context vector at time point t, which is calculated by Equation (12) for capturing the visual information associated with a certain input location of the annotation vector. In Equations (8) and (9), @ stands for the element-wise multiplication. In Equation (10), f att is the attention function which is modeled by a multilayer perceptron to generate the weight e ti based on the annotation vector a i and the output h t−1 of previous time step.…”

Section: Tsmc-cnn-alstmmentioning

confidence: 99%

Time Series Multiple Channel Convolutional Neural Network with Attention-Based Long Short-Term Memory for Predicting Bearing Remaining Useful Life

Jiang

Lee

Zeng

2019

Sensors

View full text Add to dashboard Cite

This paper proposes two deep learning methods for remaining useful life (RUL) prediction of bearings. The methods have the advantageous end-to-end property that they take raw data as input and generate the predicted RUL directly. They are TSMC-CNN, which stands for the time series multiple channel convolutional neural network, and TSMC-CNN-ALSTM, which stands for the TSMC-CNN integrated with the attention-based long short-term memory (ALSTM) network. The proposed methods divide a time series into multiple channels and take advantage of the convolutional neural network (CNN), the long short-term memory (LSTM) network, and the attention-based mechanism for boosting performance. The CNN performs well for extracting features from data with multiple channels; dividing a time series into multiple channels helps the CNN extract relationship among far-apart data points. The LSTM network is excellent for processing temporal data; the attention-based mechanism allows the LSTM network to focus on different features at different time steps for better prediction accuracy. PRONOSTIA bearing operation datasets are applied to the proposed methods for the purpose of performance evaluation and comparison. The comparison results show that the proposed methods outperform the others in terms of the mean absolute error (MAE) and the root mean squared error (RMSE) of RUL prediction.

show abstract

Section: Tsmc-cnn-alstmmentioning

confidence: 99%

Time Series Multiple Channel Convolutional Neural Network with Attention-Based Long Short-Term Memory for Predicting Bearing Remaining Useful Life

Jiang

Lee

Zeng

2019

Sensors

View full text Add to dashboard Cite

show abstract

“…Since the variance covariance matrix R is symmetric and non-negative definite, based on the Spectral Decomposition Theorem, there exists a symmetric and non-negative definite matrix B such that BB = R. The decomposition process is provided in Appendix. The model after transformation is shown in (6).…”

Section: B Spectral Decomposition Of Error Effect Matrixmentioning

confidence: 99%

“…As one of the important components of the train, the wheel directly determines the stability and safety of the vehicle's operating conditions and it is the core of the bogie [1]. Wear is the main failure mode of the wheel, excessive wear will make the train run unstable when passing the corner, and even cause derailment accidents [6]. The wear of wheel is a typical degradation process.…”

mentioning

confidence: 99%

Train Wheel Degradation Modeling and Remaining Useful Life Prediction Based on Mixed Effect Model Considering Dependent Measurement Errors

Zhou

Jiang

et al. 2019

IEEE Access

Self Cite

View full text Add to dashboard Cite

In order to predict the remaining useful life of rail train wheels progressively under dependent measurement errors, an extended mixed effect degradation model and Bayesian parameter update framework is proposed. The heteroscedastic and correlated structure of error terms are considered in the modeling process, and the spectral decomposition of the error effect matrix is used to realize the independent conversion of online monitoring data. Based on the Bayesian method, the online monitoring data is merged with historical knowledge for parameter updating, and remaining useful life prediction and reliability evaluation are performed. The case study results show that the proposed method considering dependent measurement errors could provide adequate degradation modeling and improve the accuracy of remaining useful life prediction. These results will help the development of prognostic and health management of train wheel. INDEX TERMS Condition monitoring, Bayesian methods, remaining life assessment, mixed effect models, degradation modeling, spectral decomposition. I. INTRODUCTION

show abstract

Assessing the performance of different devices in railway wheelset inspection

2020

View full text Add to dashboard Cite

Remaining useful life estimation of metropolitan train wheels considering measurement error

Cited by 7 publications

References 25 publications

Time Series Multiple Channel Convolutional Neural Network with Attention-Based Long Short-Term Memory for Predicting Bearing Remaining Useful Life

Time Series Multiple Channel Convolutional Neural Network with Attention-Based Long Short-Term Memory for Predicting Bearing Remaining Useful Life

Train Wheel Degradation Modeling and Remaining Useful Life Prediction Based on Mixed Effect Model Considering Dependent Measurement Errors

Assessing the performance of different devices in railway wheelset inspection

Contact Info

Product

Resources

About