Intelligent Diagnosis for Railway Wheel Flat Using Frequency-Domain Gramian Angular Field and Transfer Learning Network

Bai, Yongliang; Yang, Jianwei; Wang, Jinhai; Li, Qiang

doi:10.1109/access.2020.3000068

Cited by 29 publications

(25 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By assessing related work concerning these three dimensions, it can be seen that studies are either restricted to one dataset, belonging to one isolated domain [35], [38], only base their findings on a single visualization technique [39], or are built upon a small sample size [31], manifesting limited validity and do not allow to expand proposed approaches to classification problems with an insufficient sample size [34]. As studies have shown that using non-raw data and including manual feature extraction is highly time-consuming [17] and includes the risk of wrongfully classifying relevant features in the time-series data [12], [16], the state-of-research manifests high potential for improvement of objectivity and efficiency.…”

Section: Related Workmentioning

confidence: 99%

“…However, despite industries increasingly implementing sensor devices [5] and more than 65% of the Fortune 1000 companies already using benchmarking to obtain competitive advantages [29], [37], the current state of research lacks a systematic evaluation on how varying these time-series imaging-related dimensions influence the classification performance across domains. Studies are either adapted to a specific domain [38], only use a single visualization technique [39], manifest a small sample size [31], or manifest the theoretical issue of long-term dependencies as their approaches are based on RNNs [40], [41]. Therefore, related work is limited either theoretically [40], [41] or practically [35], [38], [39] regarding the systematic analysis of raw time-series data across domains.…”

Section: Introductionmentioning

confidence: 99%

“…Studies are either adapted to a specific domain [38], only use a single visualization technique [39], manifest a small sample size [31], or manifest the theoretical issue of long-term dependencies as their approaches are based on RNNs [40], [41]. Therefore, related work is limited either theoretically [40], [41] or practically [35], [38], [39] regarding the systematic analysis of raw time-series data across domains.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

2022

View full text Add to dashboard Cite

With the growing availability and complexity of time-series sequences, scalable and robust machine learning approaches are required that overcome the sampling challenge of quantitatively sufficient training data. Following the research trend towards the deep learning-based analysis of time-series encoded as images, this study proposes a time-series imaging workflow that overcomes the challenge of quantitatively limited sensor data across domains (i.e., medicine and engineering). After systematically identifying the three relevant dimensions that affect the performance of the deep learning-based analysis of visualized timeseries data, we performed a benchmarking evaluation with a total of 24 unique convolutional neural network models. Following a two-level transfer learning investigation, we reveal that fine-tuning the mid-level features results in the best classification performance. As a result, we present an optimized representation of the VGG16 network, which outperforms previous studies in the field. Our approach is accurate, robust, and manifests internal and external validity. By only using the raw time-series data, our model does not require manual feature engineering, being of high practical relevance. As the post-hoc analysis of our results reveals that our model allows automated extraction of meaningful features based on the trend of the underlying timeseries data, our study also adds to explainable artificial intelligence. Furthermore, our proposed workflow reduces the sequence length of the input data while preserving all information. Especially with the hurdle of long-term dependencies in sequential time-series data, we overcome related work's limitation of the vanishing gradients problem and contribute to the sequential learning theory in artificial intelligence.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

2022

View full text Add to dashboard Cite

show abstract

“…In [17], a data driven approach to estimate the length of a wheel flat is proposed. Bai et al [18] presented a frequency-domain image classification method to analyze wheel flats.…”

Section: Introductionmentioning

confidence: 99%

Train Wheel Condition Monitoring via Cepstral Analysis of Axle Box Accelerations

et al. 2021

View full text Add to dashboard Cite

Continuous wheel condition monitoring is indispensable for the early detection of wheel defects. In this paper, we provide an approach based on cepstral analysis of axle-box accelerations (ABA). It is applied to the data in the spatial domain, which is why we introduce a new data representation called navewumber domain. In this domain, the wheel circumference and hence the wear of the wheel can be monitored. Furthermore, the amplitudes of peaks in the navewumber domain indicate the severity of possible wheel defects. We demonstrate our approach on simple synthetic data and real data gathered with an on-board multi-sensor system. The speed information obtained from fusing global navigation satellite system (GNSS) and inertial measurement unit (IMU) data is used to transform the data from time to space. The data acquisition was performed with a measurement train under normal operating conditions in the mainline railway network of Austria. We can show that our approach provides robust features that can be used for on-board wheel condition monitoring. Therefore, it enables further advances in the field of condition based and predictive maintenance of railway wheels.

show abstract

“…Chen, Wang 18 presented that fault diagnosis should be changed from behavioral research to mechanism research, and emphasized the importance of dynamic analysis and signal processing. Although PHM technology including reliability evaluation 19 and fault diagnosis 20 can greatly improve the safety of railway vehicles, dynamic modeling is still the basis of health management of railway vehicles for better understanding the fault mechanism. 21 The accurate and straightforward fault modeling and effective signal processing method is the key to realize the vibration feature analysis of faults of axlebox bearing of a railway vehicle.…”

Section: Introductionmentioning

confidence: 99%

A comparative study of the vibration characteristics of railway vehicle axlebox bearings with inner/outer race faults

Wang

Yang

Bai

et al. 2020

Proceedings of the Institution of Mechanical Engineers, Part F:

Self Cite

View full text Add to dashboard Cite

Increasing service time makes the axlebox bearing of railway vehicle vulnerable to develop a fault in inner or outer races, which can cause some serious adverse effects on a railway vehicle’s safe operation. To tackle this problem, we established a railway vehicle vertical-longitudinal dynamic model with inner/outer races faults of axlebox bearing and validated it by experimental data. We utilized the time-synchronous average (TSA) technology to filter the raw signals and studied their vibration features. The results show that the longitudinal vibration features are more sensitive for inner race fault identification, while the vertical vibration features are more suitable for outer race fault identification. For inner race fault identification, the indicator peak-to-peak value (PPV) that increases 1056% relative to the healthy state at the most severe fault performs the best sensitivity. For outer race fault identification, the indicator skewness value (SV) that increases 518% relative to the healthy state at the most severe fault exhibits the best performance. The research work can provide meaningful guidance for accurate diagnosis of axlebox bearing faults of railway vehicles.

show abstract

Intelligent Diagnosis for Railway Wheel Flat Using Frequency-Domain Gramian Angular Field and Transfer Learning Network

Cited by 29 publications

References 29 publications

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

Train Wheel Condition Monitoring via Cepstral Analysis of Axle Box Accelerations

A comparative study of the vibration characteristics of railway vehicle axlebox bearings with inner/outer race faults

Contact Info

Product

Resources

About