Wheel fault diagnosis model based on multichannel attention and supervised contrastive learning

As wheels are important components of train operation, diagnosing and predicting wheel faults are essential to ensure the reliability of rail transit. Currently, the existing studies always separately deal with two main types of wheel faults, namely wheel radius difference and wheel flat, even though they are both reflected by wheel radius changes. Moreover, traditional diagnostic methods, such as mechanical methods or a combination of data analysis methods, have limited abilities to efficiently extract data features. Deep learning models have become useful tools to automatically learn features from raw vibration signals. However, research on improving the feature-learning capabilities of models under noise interference to yield higher wheel diagnostic accuracies has not yet been conducted. In this paper, a unified training framework with the same model architecture and loss function is established for two homologous wheel faults. After selecting deep residual networks (ResNets) as the backbone network to build the model, we add the squeeze and excitation (SE) module based on a multichannel attention mechanism to the backbone network to learn the global relationships among feature channels. Then the influence of noise interference features is reduced while the extraction of useful information features is enhanced, leading to the improved feature-learning ability of ResNet. To further obtain effective feature representation using the model, we introduce supervised contrastive loss (SCL) on the basis of ResNet + SE to enlarge the feature distances of different fault classes through a comparison between positive and negative examples under label supervision to obtain a better class differentiation and higher diagnostic accuracy. We also complete a regression task to predict the fault degrees of wheel radius difference and wheel flat without changing the network architecture. The extensive experimental results show that the proposed model has a high accuracy in diagnosing and predicting two types of wheel faults.

Flexible-Rigid Wheelset Introduced Dynamic Effects due to Wheel Tread Flat

The widespread faults that occur in railway wheels and can cause a massive dynamic impact are the wheel tread flat. The current work considered changes in vehicle speed or wheel radius deviation and studied the dynamic impact load. The modal technique for the impact evaluation induced by the wheel flat was proposed via the finite element analysis (FEA) software package ANSYS, integrated into a multibody dynamics model of the high-speed train CRH2A (EMU) through SIMPACK. The irregularity track line has developed and depends on the selected simulation data points. Additionally, a statistical approach is designed to analyze the dynamic impact load response and effect and consider different wheel flat lengths and vehicle speeds. The train speed influence on the flat size of the vertical wheel-rail impact response and the statistical approach are discussed based on flexible, rigid wheelsets. The results show that the rigid wheel flat has the highest vertical wheel impact load and is more significant than the flexible wheel flat force. The consequences suggest that the wheelset flexibility can significantly improve vertical acceleration comparably to the rigid wheel flats. In addition, the rendering of the statistical approach shows that the hazard rate, PDF, and CDF influence increase when the flat wheel length increases.

Dynamic Influence of Wheel Flat on Fatigue Life of the Traction Motor Bearing in Vibration Environment of a Locomotive

Wheel flat can cause a large impact between the wheel and rail and excites a forced vibration in the locomotive and track structure systems. The working conditions and fatigue life of the motor bearings are significantly affected by the intensified wheel–rail interaction via the transmission path of the gear mesh. In this study, a fatigue life prediction method of the traction motor bearings in a locomotive is proposed. Based on the L−P theory or ISO 281 combined with the Miner linear damage theory and vehicle–track coupled dynamics, the irregular loads induced by the track random irregularity and gear mesh are considered in this proposed method. It can greatly increase the accuracy of predictions compared with the traditional prediction models of a rolling bearing life whose bearing loads are assumed to be constant. The results indicate that the periodic impact forces and larger mesh forces caused by the wheel flat will reduce the fatigue life of the motor bearings, especially when the flat length is larger than 30 mm. Using this method, the effects of the flat length and relatively constant velocity of the locomotive are analyzed. The proposed method can provide a theoretical basis to guarantee safe and reliable working for motor bearings.

Wheel Flats in the Dynamic Behavior of Ballasted and Slab Railway Tracks

Wheel flats induce high-impact loads with relevance for the safety of the vehicle in operation as they can contribute to broken axles, hot axle boxes, and damaged rolling bearings and wheels. The high loads also induce damage in the track components such as rails and sleepers. Although this subject has been studied numerically and experimentally over the last few years, the wheel flat problem has focused on ballasted tracks, and there is a need to understand the phenomena also for slab tracks. In this research, a numerical approach was used to show the effects of the wheel flats with different geometric configurations on the dynamic behavior of a classical ballasted track and a continuous slab track. Several wheel flat geometries and different vehicle speeds were considered. The nonlinear Hertzian contact model was used because of the high dynamic variation of the interaction of the load between the vehicle and the rail. The results evidenced that, for the same traffic conditions, the dynamic force was higher on the slab track than on the ballasted one, contrary to the maximum vertical displacement, which was higher on the ballasted track due to the track differences regarding the stiffness and frequency response. The results are useful for railway managers who wish to monitor track deterioration under the regulatory limits.

Determination of mapping relation between wheel flat and wheel/rail contact force for railway freight wagon using dynamic simulation

In order to comprehensively consider the dynamic behavior of vehicle system and the contact forces between wheel and rail, the vehicle-track coupling model is established considering the flexible wheelset and rail modal characteristics. Guyan reduction theory is introduced to reduce the degrees of freedom of wheelset and rail and to improve the calculation speed. Due to small axle load variation of freight wagon operating on the special line for the coal transportation, the vehicles with different speeds and wheel flat lengths, whether in empty or loaded conditions, are simulated and then the mapping relations between the flat lengths and wheel/rail impact force are obtained. Subsequently, the fitting functions for the empty and loaded vehicles are fitted to quantitatively detect the wheel flat by trackside equipment. The results indicate that the fitting function of empty vehicle has a better effect on predicting the flat length within 6% error since wheel/rail contact force of empty vehicle induced by wheel flat increases with the increase of flat length, while that of loaded vehicle presents a parabola variation trend at low speed and increasing trend at higher speed.

Wheel flat detection and severity classification using deep learning techniques

Wheel flats are one of the most common types of defect found in railway systems. Wheel flats can result in decreasing passenger comfort and noise if they are slight, or serious incidents such as derailment if they are severe. With the increasing demand for railway transport, the speed and weight of rolling stock tend to increase, which results in relatively rapid deterioration. The occurrence of wheel flats is also affected by this increasing demand. To perform preventative maintenance for wheel flats, to keep wheelsets in a proper condition and to minimise maintenance costs, the ability to detect and classify wheel flats is required. This study aims to apply deep learning techniques to detect wheel flats and classify wheel flat severity. The deep learning techniques used in the study are a deep neural network (DNN), a convolutional neural network (CNN) and a recurrent neural network (RNN). 1608 samples, simulated using D-Track, a dynamic behaviour simulation software package, are used to develop machine learning models. Three different aspects of the models are evaluated, namely overall accuracy, the ability to detect wheel flats and the ability to classify wheel flat severity. The results from the study show the DNN has the highest overall accuracy of 96%. In addition, the DNN can be used to detect wheel flats with nearly 100% accuracy. The CNN performs better than the RNN in terms of overall accuracy and wheel flat detection. However, the RNN performs better than the CNN in wheel flat severity classification. Overall, the DNN offers the best approach for detecting wheel flats and classifying their severity.

Railway Vehicle Wheel Flat Detection with Multiple Records Using Spectral Kurtosis Analysis

The gradual deterioration of train wheels can increase the risk of failure and lead to a higher rate of track deterioration, resulting in less reliable railway systems with higher maintenance costs. Early detection of potential wheel damages allows railway infrastructure managers to control railway operators, leading to lower infrastructure maintenance costs. This study focuses on identifying the type of sensors that can be adopted in a wayside monitoring system for wheel flat detection, as well as their optimal position. The study relies on a 3D numerical simulation of the train-track dynamic response to the presence of wheel flats. The shear and acceleration measurement points were defined in order to examine the sensitivity of the layout schemes not only to the type of sensors (strain gauge and accelerometer) but also to the position where they are installed. By considering the shear and accelerations evaluated in 19 positions of the track as inputs, the wheel flat was identified by the envelope spectrum approach using spectral kurtosis analysis. The influence of the type of sensors and their location on the accuracy of the wheel flat detection system is analyzed. Two types of trains were considered, namely the Alfa Pendular passenger vehicle and a freight wagon.