# Real-Time Hybrid Deep Learning-Based Train Running Safety Prediction Framework of Railway Vehicle

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Background and Literature Review

## 3. Results Train Running Safety Data and Measurement Framework

## 4. Real-Time Deep-Learning-Based Train Running Safety Prediction Framework

## 5. Verification and Analysis of Hybrid Deep-Learning Prediction Framework for Train Running Safety

## 6. Conclusions and Further Study

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Conflicts of Interest

## References

- UIC. Testing and Approval of Railway Vehicles from the Point of View of Their Dynamic Behaviour-Safety-Track Fatigue-Ride Quality, 4th ed.; International Union of Railways: Paris, France, 2009. [Google Scholar]
- BSI. BS EN 14067-1, Railway Applications- Aerodynamics-Part1: Symbols and Units, 3rd ed.; British Standard Institution: London, UK, 2011. [Google Scholar]
- BSI. BS EN 14067-6:2018-TC Railway Applications-Aerodynamics: Requirements and Test Procedures for Cross Wind Assessment, 1st ed.; British Standard Institution: London, UK, 2020. [Google Scholar]
- KRRI. KRTS-VE-Part31-2014(R1) Technical Specifications for High Speed Railway Vehicles, 1st ed.; Korea Railroad Research Institute: Uiwang, Korea, 2014. [Google Scholar]
- KRRI. KRTS-VE-Part21-2015(R1) Technical Specifications for High Speed Railway Vehicles, 1st ed.; Korea Railroad Research Institute: Uiwang, Korea, 2015. [Google Scholar]
- BSI. EN 14363:2016 Railway Applications—Testing and Simulation for the Acceptance of Running Characteristics of Railway Vehicles-Running Behavior and Stationary Tests, 1st ed.; British Standard Institution: London, UK, 2016. [Google Scholar]
- UIC. UIC Code 518 OR Testing and Approval of Railway Vehicles from the Point of View of Their Dynamic Behavior-Safety-Track Fatigue-Ride Quality, 1st ed.; Worldwide Railway Organisation: Paris, France, 2003. [Google Scholar]
- KRRI. KRTS-VE-Part51-2017(R1) Technical Specifications for High Speed Railway Vehicles, 1st ed.; Korea Railroad Research Institute: Uiwang, Korea, 2017. [Google Scholar]
- Arvidsson, T.; Andersson, C.; Karoumi, R. Train running safety on non-ballasted bridges. Int. J. Rail Transp.
**2018**, 7, 1–22. [Google Scholar] [CrossRef] [Green Version] - Diang, Y.; Sun, P.; Wang, G.; Song, Y.; Wu, L.; Yue, Q.; Li, A. Early-warning method of train running safety of a high-speed railway bridge based on transverse vibration monitoring. Shock Vib.
**2015**, 2015, 1–9. [Google Scholar] [CrossRef] [Green Version] - Choi, J.; Kim, J.; Chung, J.; Lee, S. Evaluation of Training running safety for direct fixation concrete track on light rapid transit. J. Korean Soc. Saf.
**2017**, 32, 41–46. [Google Scholar] - Jang, S.; Yang, S. Assessment of train running safety, ride comfort and track serviceability at transition between floating slab track and conventional concrete track. J. Korean Soc. Railw.
**2012**, 15, 48–61. [Google Scholar] [CrossRef] [Green Version] - Kim, M.K.; Eom, B.G.; Lee, H.S. Running Safety Analysis of Railway Vehicle Passing through Curve Depending on Rail Inclination Change. Korean Soc. Noise Vib. Eng.
**2013**, 23, 199–208. [Google Scholar] [CrossRef] [Green Version] - Oh, J.T.; Kwon, T.S. A Study on the Assessment of Derailment Factor for the Enhancement of Train Running Safety. In Proceedings of the Spring Conference & Annual Meeting of the Korean Society for Railway, Changwon, Korea, 6 October 2000; Volume 2000, pp. 210–217. [Google Scholar]
- Seo, S.; Park, J.H.; Min, S.H. Studies on Safety Criteria for Trains Running on Floating Railway Bridges. Advances in Structural Engineering. Available online: https://journals.sagepub.com/doi/abs/10.1177/1369433220980524 (accessed on 30 March 2021).
- Zhang, X.; Zhou, S.; Di, H.; He, C. A semi-analytical model of the train-floating slab track-tunnel-soil system considering the non-nonlinear wheel/rail contact. J. Rail Rapid Transit
**2018**, 232, 2063–2078. [Google Scholar] [CrossRef] - Alawad, H.; Kaewunruen, S.; An, M. A deep learning approach towards railway safety risk assessment. IEEE Access
**2020**, 8, 102811–102832. [Google Scholar] [CrossRef] - Yang, C.; Sun, Y.; Ladubec, C.; Liu, Y. Developing machine learning-based models for railway inspection. Appl. Sci.
**2021**, 11, 1–15. [Google Scholar] - He, K.; Zhang, X.; Ren, S.; Sun, J. Deep residual learning for image recognition. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27–30 June 2016; pp. 1063–6919. [Google Scholar]
- Lee, H.; Han, S.; Park, K. Generative adversarial network-based missing data handling and remaining useful life estimation for smart train control and monitoring systems. J. Adv. Transp.
**2020**, 2020, 1–15. [Google Scholar] - Vampire Pro. Available online: https://www.ensco.com/rail/vampire (accessed on 10 January 2021).
- Cherchas, D.B. Determination of railway wheel climb probability based on the derailment coefficient. J. Frankl. Inst.
**1981**, 312, 31–40. [Google Scholar] [CrossRef] - Wang, P.; Wang, J.; Ma, X.; Ma, D.; Xu, J.; Q, Y. Theoretical 3D model for Quasistatic critical derailment coefficient of railway vehicles and a simplified formula. Math. Probl. Eng.
**2017**, 2018, 1–14. [Google Scholar] [CrossRef] [Green Version] - Wang, K.; Huang, C.; Zhai, W.; Liu, P.; Wang, S. Progress on wheel-rail dynamic performance of railway curve negotiation. J. Traffic Transp. Eng.
**2014**, 1, 209–220. [Google Scholar] [CrossRef] [Green Version] - Vollebregt, E. Detailed wheel/rail geometry processing with the conformal contact approach. Multibody Syst. Dyn.
**2020**, 2020, 1–33. [Google Scholar] [CrossRef] - Zhou, L.; Brunskill, H.; Pletz, M.; Daves, W.; Scheriau, S.; Lewis, R. Real-time measurement of dynamic wheel-rail contacts using ultrasonic reflectometry. J. Tribol.
**2019**, 141, 1–9. [Google Scholar] [CrossRef] - Jo, Y.; Lee, H. Electricity demand forecasting framework using modified attention-based LSTM. J. Korean Inst. Intell. Syst.
**2020**, 30, 242–250. [Google Scholar] [CrossRef] - Zhang, Q.; Zhuang, Y.; Wei, Y.; Jiang, H.; Yang, H. Railway safety risk assessment and control optimization method based on FTA-FPN: A case study of Chinese high-speed railway station. J. Adv. Transp.
**2020**, 2020, 1–11. [Google Scholar] [CrossRef] - Bae, H.; Yun, K.; Moon, J.; Lim, N. Impact force evaluation of the derailment containment wall for high-speed train through a collision simulation. Adv. Civ. Eng.
**2018**, 2018, 1–14. [Google Scholar] [CrossRef] [Green Version] - Rahmadani, F.; Lee, H. Hybrid deep learning-based epidemic prediction framework of COVID-19: South Korea case. Appl. Sci.
**2020**, 10, 8539. [Google Scholar] [CrossRef]

**Figure 2.**Allowance of dynamic running safety: (

**a**) Allowance per derailment coefficient (DC); (

**b**) allowance of running safety per dynamic vertical power (DV).

**Figure 5.**The target railway and its rail model: (

**a**) The target real railway; (

**b**) track distance plot of the modelled railway.

**Figure 6.**Vibrations from the transient analysis using the modeled railway and HEMU-430X: (

**a**) Vibration plot; (

**b**) model data and output data using the transient analysis.

**Figure 7.**Statistical correlation test among 28 attributes. (

**a**) Correlation test between attributes. (

**b**) Correlation between “Railway point” and “Right wheel derail coefficient (DC)”.

**Figure 8.**Data plot of the “Wheel lateral pressure” from a transient analysis: (

**a**) Data plot of wheel lateral pressure; (

**b**) standard deviation plot of wheel lateral pressure.

**Figure 9.**A general deep neural network architecture and stationary characteristics of output variables. (

**a**) A general deep neural network architecture. (

**b**) Stationary characteristics of output variables in train running safety.

**Figure 11.**Comparisons between both frameworks: (

**a**) RMSE using the DNN model; (

**b**) loss using the DNN model; (

**c**) RMSE using the proposed hybrid deep learning framework; (

**d**) loss using the proposed hybrid deep learning framework.

**Figure 12.**Prediction results using the test set: (

**a**) Prediction results using the DNN model; (

**b**) prediction results using the DNN model.

Symbol | Terms | Unit |
---|---|---|

L | Lateral force | kN |

V | Vertical force | kN |

N | Normal force | kN |

$\mathsf{\alpha}$ | Contact angle (Flange contact angle) | $\xb0$ |

${\mathrm{T}}_{\mathrm{Y}}$ | Tangential force | kN |

Y | Lateral force per a wheel axis | kN |

P | Axle load | kN |

$\mathsf{\mu}$ | Friction coefficient | $\mathsf{\mu}\in \mathrm{R}$ (R is real number) |

$\mathsf{\Delta}\mathrm{V}$ | Gap between consecutive vertical forces | kN |

Classification | Measurements | Unit | Criteria |
---|---|---|---|

Train running safety | Rate of wheel load reduction (DV) | $\mathrm{R}\in \left[0,1\right]$, R is a real number | $\mathrm{DV}\le 0.13$ |

Derailment coefficient (DC) | R | $\mathrm{DC}\le 0.8$ | |

Lateral displacement of rail head (LD) | mm | $\mathrm{LD}\le 4$ |

Existing Research Studies | Characteristics | Used Methods | Issues |
---|---|---|---|

Arvidsson et al. [9] | - Running safety simulation under non-ballasted bridge environments - Simulation analysis of running safety and passenger comport | - Simulation using 2D train-track-bridge model | - Predefined model-based simulation studies |

Ding, et al. [10] | - Early warning framework with vibrations of an express train | - Nonlinear equation-based regress model | - Monitoring-based early warning framework |

Choi, et al. [11] | - Light rail (LRT)-based vibration measurement on real running environment | - Real measurement | - Limited in small distance-measurement |

Jang and Yang [12] | - Numerical simulation—Consideration on transition between floating slab track and concrete track | - DIASTARS-based CAE simulation | - Limited experimental condition |

Kim, et al. [13] | - CAE-based simulation studies | - Input of “real railway models and conditions” | - CAE-based analysis |

Oh and Kwon [14] | - Measure on real train - Exemplary proof of DV’s importance on running safety | - Vibration measurement on trains with different weights | - Single factor (weight)-based experiment |

Seo, et al. [15] | - Simulation study- Relationship between train wheels and floating railway bridges | - Modeling of floating railway bridges - Nonlinear equation-based wheel motion model | - Nonlinear equation-based simulation model |

Zhang, et al. [16] | - 3D simulation model of train-induced vibration of a floating slab | - Train/environment model-based simulation | - Model-based simulation study |

Classification | Attribute | Unit | Data Source | |
---|---|---|---|---|

Modeling Input Form Real Measurement | Generation Using Transient Analysis | |||

Railway model data | Railway point (distance) | mm | O | - |

Cross level irregularity (cant) | mm | O | - | |

Curvature irregularity | 1/km | O | - | |

Lateral irregularity | mm | O | - | |

Vertical irregularity | mm | O | - | |

Gauge variation | mm | O | - | |

Train structure/ simulation data | Bogie upper frame lateral vibration | $\mathrm{m}/{\mathrm{s}}^{2}$ | - | O |

Bogie upper frame vertical vibration | $\mathrm{m}/{\mathrm{s}}^{2}$ | - | O | |

Bogie upper body lateral vibration | $\mathrm{m}/{\mathrm{s}}^{2}$ | - | O | |

Bogie upper body vertical vibration | $\mathrm{m}/{\mathrm{s}}^{2}$ | - | O | |

Left wheel lateral weight | kg | - | O | |

Right wheel lateral weight | kg | - | O | |

Left wheel vertical weight | kg | - | O | |

Right wheel vertical weight | kg | - | O | |

Left wheel derail coefficient (DC) | Real number | - | O | |

Right wheel derail coefficient (DC) | Real number | - | O | |

Left wheel rate of load reduction (DV) | Real number | - | O | |

Right wheel rate of load reduction (DV) | Real number | - | O | |

Body frame lateral pressure (body frame lateral forces) | kN | - | O | |

Left axle box lateral vibration | $\mathrm{m}/{\mathrm{s}}^{2}$ | - | O | |

Right axle box lateral vibration | $\mathrm{m}/{\mathrm{s}}^{2}$ | - | O | |

Left axle box vertical vibration | $\mathrm{m}/{\mathrm{s}}^{2}$ | - | O | |

Right axle box vertical vibration | $\mathrm{m}/{\mathrm{s}}^{2}$ | - | O | |

Wheel lateral pressure wheel lateral forces) | kN | - | O |

Attribute (Railway Model Parameters) | Relationships with Train Structure and Mechanism |
---|---|

Railway point (distance) | - Little relationship (r* < 0.01) |

Cant | - Weak relationship: Left axle box vertical vibration - little relationship with the other factors |

Curvature irregularity | - Little relationship (r* < 0.01) |

Lateral irregularity | - Strong relationship: Bogie upper frame lateral vibration - Weak relationship: Right wheel lateral weight, Right wheel DV, wheel lateral pressure - little relationship with the other factors |

Vertical irregularity | - Strong relationship: Bogie upper frame vertical vibration - Weak relationship: Left/right wheel vertical weight, Left/right wheel DC, Left/right axle box vertical vibration - little relationship with the other factors |

Gauge variation | - Weak relationship: Wheel lateral pressure - little relationship with the other factors |

Classification | A DNN without Recurrent Data | The Proposed Hybrid Network |
---|---|---|

Input | ${\mathrm{X}}_{\mathrm{i},\mathrm{i}\in \mathrm{N}\left[1,6\right]}\left(t\right)$, ${\mathrm{X}}_{\mathrm{i},\mathrm{i}\in \mathrm{N}\left[7,18\right]}\left(t\right)$ | ${\mathrm{X}}_{\mathrm{i},\mathrm{i}\in \mathrm{N}\left[1,18\right]}\left(t\right),$ ${\mathrm{X}}_{\mathrm{i},\mathrm{i}\in \mathrm{N}\left[7,18\right]}\left(t\right)$ ${\widehat{Y}}_{j}{\left(t-k\cdot \mathsf{\Delta}t\right)}_{j\in N\left[1,5\right]},k\in N\left[1,5\right]$ |

Output | ${\widehat{Y}}_{j}{\left(t\right)}_{j\in N\left[1,5\right]}$ | |

Layer architecture | 4 hidden layers Number of hidden nodes in each hidden layer = {40,30,15,5} | 4 hidden layers Number of hidden nodes in each hidden layer = {50,30,15,5} |

Activation functions | Sigmoid/ReLU Sigmoid: $\frac{1}{1+{\mathrm{e}}^{-\mathrm{x}}}$ RelU: max(0,x) | |

Learning parameters | Epoch = 5000/optimization method = ADAM () $\mathrm{Learning}\mathrm{rate}(\mathsf{\eta})$= 0.001 Dropout rate = 0.2 |

Classification | LSTM | DNN Using | The Proposed Framework |
---|---|---|---|

Input | ${\widehat{Y}}_{j}{\left(t-k\cdot \mathsf{\Delta}t\right)}_{j\in N\left[1,5\right]}$,$k\in N\left[1,5\right]$ | Refer Table 5 | Refer Table 5 |

Output | ${\widehat{Y}}_{j}{\left(t\right)}_{j\in N\left[1,5\right]}$ | ||

Parameters | Number of hidden dimensions = 20 State activation function = sigmoid Gate activation function = tanh Learning rate ($\mathsf{\eta}$) = 0.001 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Lee, H.; Han, S.-Y.; Park, K.; Lee, H.; Kwon, T.
Real-Time Hybrid Deep Learning-Based Train Running Safety Prediction Framework of Railway Vehicle. *Machines* **2021**, *9*, 130.
https://doi.org/10.3390/machines9070130

**AMA Style**

Lee H, Han S-Y, Park K, Lee H, Kwon T.
Real-Time Hybrid Deep Learning-Based Train Running Safety Prediction Framework of Railway Vehicle. *Machines*. 2021; 9(7):130.
https://doi.org/10.3390/machines9070130

**Chicago/Turabian Style**

Lee, Hyunsoo, Seok-Youn Han, Keejun Park, Hoyoung Lee, and Taesoo Kwon.
2021. "Real-Time Hybrid Deep Learning-Based Train Running Safety Prediction Framework of Railway Vehicle" *Machines* 9, no. 7: 130.
https://doi.org/10.3390/machines9070130