Training Speed Research Articles

Exploring optimizer efficiency for facial expression recognition with convolutional neural networks

AbstractIt's widely accepted that human expressions, considering for roughly sixty percent of all daily interactions, are among the most authentic forms of communication. Numerous studies are being conducted to explore the importance of facial expressions and the development of machine‐assisted recognition techniques. Significant progress is being made in facial and expression recognition, largely due to the rapid growth of machine learning and computer vision. A variety of algorithmic approaches and methods exist for detecting and recognizing facial expressions and features. This study investigates various optimization algorithms used with convolutional neural networks for facial expression recognition. The primary focus is on Adam, RMSProp, stochastic gradient descent and AdaMax optimizers. A comprehensive comparison is being made, examining the key aspects of each optimizer, including its advantages and disadvantages. Furthermore, the study also incorporates findings from recent studies that used these optimizers in various applications, highlighting their performance in terms of training time and precision. The aim is to illuminate the process of selecting a suitable optimizer for specific applications, analysing the trade‐offs between training speed and higher accuracy levels. Moreover, this study provides a deeper analysis of the role optimizers play in machine learning‐based facial expression recognition models. The discussion of the technical challenges posed by these optimizers and future improvements for achieving much more optimal results concludes the study.

Investigation of the stability of the movement of mainline road trains with semi-trailers having a rotary controlled undercarriage module

In the modern world, the demand for highway transportation involving special vehicles is increasing every year. To solve such problems, two-link road trains are widely used due to their increased load capacity, cross-country ability and large dimensions of trailer and semi-trailer links, which make it possible to solve transport problems most effectively. When designing mainline road trains, they strive to maximize the length of trailed links with pivoting undercarriage modules. In this regard, a number of difficulties arise due to the need to improve the operational properties of road trains, such as maneuverability, controllability and stability of movement. However, despite the fact that semi-trailers with pivoting undercarriage modules are being manufactured and are in operation, the theoretical issues related to their stability and controllability in various driving conditions have not been sufficiently worked out. The purpose of the work is to study the stability of the movement of mainline road trains with semi-trailers having a rotary controlled undercarriage module. A study was conducted on the risks of loss of stability of mainline road trains with semi-trailers having a rotary controlled undercarriage module in various operating modes. It has been established that in order to ensure stable movement of the trailer link with a rotary support module, it is necessary to introduce corrective steering of the trolley in order to maintain the calculated folding angle of the road train and limit the angular folding speed of the road train. Keywords Stability, controllability, road train, semi-trailer, undercarriage module

Quantitative identification of wire rope core conveyor belt damage based on GWO-BP

Abstract To address the challenges of detecting internal damage in steel wire rope core conveyors and the difficulties in quantitative identification, this study proposes an improved backpropagation (BP) neural network damage identification algorithm based on the Grey Wolf Optimization (GWO-BP). The Grey Wolf algorithm is employed to optimize the initial weights and thresholds of the BP neural network, thereby enhancing its performance and stability. A testing apparatus for detecting damage in steel wire rope core conveyors is designed and constructed to evaluate the algorithm's effectiveness and feasibility. First, damage signal data from the steel wire rope are extracted and normalized to facilitate the convergence of the predictive model. Next, the BP neural network algorithm is optimized to address issues such as parameter selection randomness, improving model training speed and identification accuracy. Experimental results indicate that the optimized BP algorithm achieves an average identification accuracy of 96.0%, representing a 5.5% improvement over the unoptimized BP algorithm and significantly enhancing the precision of damage quantitative identification.

Analysis of Badminton Techniques Smash Using a Biomechanical Approach: Systematic Literature Review

Study purpose. The purpose of this research is to look into the use Analysis of Badminton Techniques Smash Using A Biomechanical Approach. Materials and methods. The PRISMA standards for systematic reviews and meta-analyses were followed in this review investigation. The study must be published with in the prior four years, from January 2020 to Desember 2024. In the search procedure, the following keywords are used: (1) Badminton; (2) Smash. Scopus search engine was used in this research. Results. The theme of this research as a whole obtained 155. Consists of 37 articles which are then taken 10 relevant articles. Conclusions. Analysis of badminton smash techniques using a biomechanical approach includes the effect of smash techniques with a biomechanical approach based on relevant results includes improving smash performance, strength and speed training, muscle strength biomechanical, improving smash speed, glutamine accelerates athlete recovery, and fatigue reduces jumping smash performance

Limb motion mechanics analysis of Taijiquan trainees combined with APAS image analysis system

Analyzing the motion mechanics of Taijiquan trainees’ limbs can help trainees better understand the mechanics of Taijiquan movement and improve their training speed. However, all the current motion mechanics analysis models still have the disadvantage of poor motion mechanics analysis due to inaccurate motion image feature extraction. Therefore, this study utilizes the Ariel performance analysis system and temporal-difference to construct a motion mechanics analysis model to extract motion image features of Taijiquan and analyze its motion mechanics. analysis. The study first experimented with this analysis model. The outcomes indicated that the model achieved 100% accuracy in the extraction of image feature information and 100% accuracy in the analysis of motion mechanics, which was much higher than the comparative analysis model. The model was then used to analyze the motion mechanics of Taijiquan trainees during bending and squatting posture. The results revealed that the larger the knee fixation angle was, the shorter the completion time of the movement was. When the knee fixation angle was 150°, the time taken to complete the starting, fixing and finishing phases of the movement was 11.21 s, 61.21 s and 12.32 s respectively. Moreover, when the trainees performed the movement, the angle of the trainee’s right knee changed gently in the starting phase, while the angle of the trainee’s left knee changed more drastically. In summary, the motion mechanics analysis model proposed in the study is able to accurately analyze the limb motion mechanics of Taijiquan trainees as a means of avoiding various injuries during training.

Fault diagnosis of nonlinear analog circuits using generalized frequency response function and LSSVM.

A fault diagnosis method of nonlinear analog circuits is proposed that combines the generalized frequency response function (GFRF) and the simplified least squares support vector machine (LSSVM). In this study, the harmonic signal is used as an input to estimate the GFRFs. To improve the estimation accuracy, the GFRFs of an analog circuit are solved directly using time-domain data. The Fourier transform of the time-domain data is avoided. After obtaining the fault features, a multi-fault classifier is designed based on the LSSVM. In order to improve the training speed and reduces storage, a simplified LSSVM model is used to construct the classifier, and the conjugate gradient algorithm is used for training. The fault diagnosis simulation experiment is conducted on a biquad filter circuit to verify the proposed method. The experimental results show that the proposed method has high diagnostic accuracy and short training time.

Exploring spiking neural networks for deep reinforcement learning in robotic tasks

Spiking Neural Networks (SNNs) stand as the third generation of Artificial Neural Networks (ANNs), mirroring the functionality of the mammalian brain more closely than their predecessors. Their computational units, spiking neurons, characterized by Ordinary Differential Equations (ODEs), allow for dynamic system representation, with spikes serving as the medium for asynchronous communication among neurons. Due to their inherent ability to capture input dynamics, SNNs hold great promise for deep networks in Reinforcement Learning (RL) tasks. Deep RL (DRL), and in particular Proximal Policy Optimization (PPO) has been proven to be valuable for training robots due to the difficulty in creating comprehensive offline datasets that capture all environmental features. DRL combined with SNNs offers a compelling solution for tasks characterized by temporal complexity. In this work, we study the effectiveness of SNNs on DRL tasks leveraging a novel framework we developed for training SNNs with PPO in the Isaac Gym simulator implemented using the skrl library. Thanks to its significantly faster training speed compared to available SNN DRL tools, the framework allowed us to: (i) Perform an effective exploration of SNN configurations for DRL robotic tasks; (ii) Compare SNNs and ANNs for various network configurations such as the number of layers and neurons. Our work demonstrates that in DRL tasks the optimal SNN topology has a lower number of layers than ANN and we highlight how the state-of-art SNN architectures used in complex RL tasks, such as Ant, SNNs have difficulties fully leveraging deeper layers. Finally, we applied the best topology identified thanks to our Isaac Gym-based framework on Ant-v4 benchmark running on MuJoCo simulator, exhibiting a performance improvement by a factor of 4.4× over the state-of-art SNN trained on the same task.

Characteristics of Noise Caused by Trains Passing on Urban Rail Transit Viaducts

With the large-scale construction of urban rail transit viaducts in China, the noise problem caused by trains traversing these sections has become increasingly prominent and a key technical challenge that restricts the sustainable development of rail transit. There are two main noise sources when trains pass on rail transit viaducts, namely, wheel-rail noise (WRN) and bridge-borne noise (BBN). However, most of the existing rail transit viaduct noise prediction models consider only a single noise source. In this study, a total noise prediction model incorporating both WRN and BBN was established using the finite element method (FEM), the boundary element method (BEM), and statistical energy analysis (SEA). The viaducts of Wuhan Metro Line 2 were selected as the research object, and noise tests of trains passing on the viaducts were carried out to validate the total noise prediction model. Based on the validated model, the spatial distribution characteristics and attenuation laws of the total noise were investigated, along with the influence of train speed on the total noise. The results show that the prediction model accurately simulated the total noise caused by trains passing on viaducts. When a train passed on the viaduct at a speed of 60 km/h, the total noise near the viaduct reached 88 dB(A) and decreased with the increase in the distance; at 120 m from the track centerline, the total noise decreased to less than 57 dB(A). As the distance increased, the total noise diminished across the entire frequency spectrum. Notably, low-frequency noise decayed at a slower rate than high-frequency noise. As the distance from the track centerline doubled, the total noise decreased by about 4.23 dB(A). The total noise increased with train speed. When the train speed doubled, the total noise at 30 m and 120 m from the track centerline increased by 6.32 dB(A) and 5.96 dB(A), respectively. The reason for this phenomenon is that the wheel-rail forces increase with the increase in train speed. This study will have important guiding significance and scientific value for the sustainable development of urban rail transit.

Characteristics and Control of Subway Train-Induced Environmental Vibration: A Case Study

With the widespread construction of the subway in the Chinese mainland, the environmental vibration caused by subway operation has attracted increasing attention. Train-induced environmental vibrations can cause structural deformation, uneven settlement of line foundations, and tunnel leakage, affecting the structural safety of lines and foundations. This research focuses on a segment of the Nanchang Metro Line 3, which has been chosen as the subject of investigation. A numerical model was developed to analyze the subway train-induced environmental vibration, employing the finite element method (FEM). Utilizing a numerical model, an investigation was conducted to examine the impact of train speed on the subway train-induced environmental vibration, the train-induced environmental vibration transmission characteristics were analyzed, and the control effects of vibration reduction tracks on train-induced environmental vibration were discussed. Train-induced vibration tests were also conducted on Nanchang Metro Line 3 to verify the control effects of various vibration reduction tracks. The results indicate that the subway train-induced environmental vibration rises as the train speed goes up, and the vibration peaks always appear around 63 Hz. When the train speed doubles, the Z-vibration level increases from about 5.1 dB to 5.9 dB. Subway train-induced environmental vibration shows a fluctuating decreasing trend with increasing distance from the centerline of the tunnel. The Z-vibration level reaches its maximum 4 m away from the centerline of the tunnel. Compared with the embedded sleeper, the vibration-damping fastener exhibits a vibration reduction effect of about 9 dB to 18 dB, the rubber vibration-damping pad exhibits a better vibration reduction effect of about 16 dB to 24 dB, and the steel spring floating plate exhibits the best vibration-damping effect of about 18 dB to 28 dB. The calculated Z-vibration levels are basically consistent with the measured values, indicating the accuracy of the calculated results of the control effects of the vibration reduction tracks.

An Effective Scheme to Accelerate NeRF for Web Applications Using Hash-based Caching and Precomputed Features

In recent years, 3D reconstruction and rendering technologies have become increasingly important in various web-based applications within the field of web technology. In particular, with the emergence of technologies such as WebGL and WebGPU, which enable real-time 3D content rendering in web browsers, immersive experiences and interactions on the web have been significantly enhanced. These technologies are widely used in applications such as 3D visualization of virtual products or 3D exploration of building interiors on real estate websites. Through these advancements, users can experience 3D content directly in their browsers without the need to install additional software, greatly expanding the possibilities of the web. Amidst this trend, the neural radiance field (NeRF) has garnered attention as a cutting-edge technology that improves the accuracy of 3D reconstruction and rendering. NeRF is a technique widely used in computer vision and graphics for reconstructing 3D spaces from 2D images taken from multiple viewpoints. By predicting the color and density of each pixel, NeRF captures the complex 3D structure and optical properties of a scene, enabling highly accurate 3D reconstructions. However, NeRF’s primary limitation is the time-consuming nature of both the training and inference processes. Research efforts to address this issue have focused on two key areas: optimizing network architectures and training procedures to accelerate scene learning, and improving inference speed for faster rendering. While progress has been made in enhancing training speed, challenges remain in improving the inference process. To address these limitations, we propose a two-step approach to significantly improve NeRF’s performance. First, we optimize the training phase through a multi-resolution hash encoding technique, reducing the computational complexity and speeding up the learning process. Second, we accelerate the inference phase by caching the input data of the NeRF MLP, which allows for faster rendering without sacrificing quality. Our experimental results demonstrate that this approach reduces training time by 68.42% and increases inference speed by 98.18%.

Optimization of Curve Correction in the Plan Using Computer Modeling to Improve Train Safety

Purpose. The authors intend to develop and implement innovative approaches to optimizing curve correction in the plan of railway tracks using computer modeling. This process aims to: improve traffic safety; improve the smoothness of the ride, which increases passenger comfort; eliminate speed limits on curved sections, which reduces travel time and increases railroad capacity; reduce track maintenance and rolling stock repair costs. The use of special software to simulate various scenarios of train traffic along the adjusted curves will allow us to assess the impact of the train flow structure on the speed, rolling stock dynamics, and track condition. Methodology. The study is based on the theory of mathematical modeling and numerical calculation methods. To find the best option for the reconstruction of railway sections, the line is considered as a complex system consisting of various infrastructure facilities. Due to their unsatisfactory technical condition, these facilities may limit the speed of traffic on certain sections. The solution to this problem is complicated due to the interdependence of such objects: the reduction of travel time on a particular section is not linear, and obtaining accurate results requires traction calculations taking into account various options for eliminating speed limits. Findings. A calculation algorithm for optimizing long sections of the railway plan is proposed. The algorithm allows to take into account the desired direction of shifts in the curves, which greatly facilitates the optimization process. An example of calculations based on the proposed algorithm is given, which demonstrates the effectiveness of the approach. Originality. The scientific and applied problem of optimizing the process of adjusting railway curves in the plan has been solved. This helps to improve the safety and speed of trains through the introduction of the latest computer modeling methods. Practical value. The implementation of the proposed technology for optimizing the railway plan will significantly increase the speed and safety of train traffic, as well as increase the throughput capacity of railway lines. This will be an important step for the integration of Ukrainian railways into the European transportation system.

A review on the research of noise generation mechanism and control technology in high-speed trains

Abstract Now it is widely accepted that lightweight design is one of the necessary conditions for high-speed trains. However, a lighter vehicle body and higher operating speed will produce higher noise, making it become one of the important problems restricting the development of high-speed railway. This becomes more and more obvious as train speed increases. The noise of high-speed trains mainly come from wheel-rail noise and aerodynamic noise, with sound radiation over an ultra-wide frequency range and transmission either to the vehicle body or to the surrounding environment, which seriously affects the riding comfortableness of passengers and the life quality of surrounding residents. How to reduce the noise of high-speed trains is one of the urgent problems to be solved at this stage. This paper will introduce the research progress of the generation mechanism and corresponding control technology of different noise sources in high-speed trains, discuss the outstanding problems and potential methods in noise control. Finally, we also propose possible comprehensive strategies for noise optimization, with a view to providing some useful references for related researchers and engineers.

Frequency and intensity of change of directions in German Bundesliga soccer

ABSTRACT The aim of this study was to investigate the change of direction (COD) frequencies and intensities of high-performance soccer players of the German Bundesliga independent of tactical and match context. COD data were collected from 18 German Bundesliga soccer teams (season 2016–2017; 308 fixtures) by an optical tracking system (OTS) (TRACAB). CODs were tracked using a modified algorithm and were sub-categorized by entry velocity (<3.0 m⋅s−1, 3.0–5.5 m⋅s−1, 5.5–7.0 m⋅s−1 and >7.0 m⋅s−1) and COD angle (20–59°, 60–119° and 120–180°). COD metric frequencies were compared between playing positions (goalkeepers, centre backs, full-backs, central midfielders, wide midfielders, and strikers). In general, regardless of entry velocity or COD angle, central midfielders consistently executed the highest number of COD actions during matches compared to the other playing positions. About ≈ 55% and ≈ 38% of CODs were <3.0 m⋅s−1 and <5.5 m⋅s−1, whereas ≈ 7% were >5.5 m⋅s−1. The distribution of COD angle types was ≈ 5% for 20–59°, ≈25% 60–119° and ≈ 70% for 120–180° COD angles. Our data provide insights into the COD demands of high-performance soccer in the German Bundesliga in terms of entry velocity and COD angles and their combination based on a large dataset of OTS data, which provides insights to facilitate the development of physical conditioning strategies, position-specific external load management, and multidirectional speed training with adequate test battery selection and return-to-play protocols for soccer players.

Improving Train Speed in Existing Short-Span Reinforced Concrete Railway Bridges Using Ballasted Pavement Retrofitting Method

Improving Train Speed in Existing Short-Span Reinforced Concrete Railway Bridges Using Ballasted Pavement Retrofitting Method

3D convolutional neural network based on spatial-spectral feature pictures learning for decoding motor imagery EEG signal

Non-invasive brain-computer interfaces (BCI) hold great promise in the field of neurorehabilitation. They are easy to use and do not require surgery, particularly in the area of motor imagery electroencephalography (EEG). However, motor imagery EEG signals often have a low signal-to-noise ratio and limited spatial and temporal resolution. Traditional deep neural networks typically only focus on the spatial and temporal features of EEG, resulting in relatively low decoding and accuracy rates for motor imagery tasks. To address these challenges, this paper proposes a 3D Convolutional Neural Network (P-3DCNN) decoding method that jointly learns spatial-frequency feature maps from the frequency and spatial domains of the EEG signals. First, the Welch method is used to calculate the frequency band power spectrum of the EEG, and a 2D matrix representing the spatial topology distribution of the electrodes is constructed. These spatial-frequency representations are then generated through cubic interpolation of the temporal EEG data. Next, the paper designs a 3DCNN network with 1D and 2D convolutional layers in series to optimize the convolutional kernel parameters and effectively learn the spatial-frequency features of the EEG. Batch normalization and dropout are also applied to improve the training speed and classification performance of the network. Finally, through experiments, the proposed method is compared to various classic machine learning and deep learning techniques. The results show an average decoding accuracy rate of 86.69%, surpassing other advanced networks. This demonstrates the effectiveness of our approach in decoding motor imagery EEG and offers valuable insights for the development of BCI.

Effect of Vehicle Vibration on Interior Noise of Railway Vehicles Passing Through Rail Corrugation Sections

This study involved experimental research to analyze the effect of rail corrugation on interior noise levels inside railway vehicles. Measurements taken on the Incheon Line 2 light rail indicated that the vehicle’s age and maintenance condition have minimal effects on interior noise. Although wheel wear slightly reduces interior noise, it is insufficient to address the issue of abnormal noise. The analysis of the relationship between vehicle vibrations and interior noise revealed that at 80 km/h and with a 24 mm rail corrugation wavelength, vibrations at 920 Hz in the axle box and car body increase, coinciding with the dominant interior noise frequency of 926.6 Hz. Furthermore, an analysis using a car body sweep confirmed a relative increase in noise in the 920 Hz range. Therefore, abnormal noise in rail corrugation sections is caused by vibrations at 920 Hz due to the corrugation wavelength and train speed, which align with the car body resonance frequency, leading to increased car body vibrations and interior noise.

Research and Simulation Application of Automatic Generation of Judging Curves for IoT Mobile Devices in Train Driving Operations Incorporating Hybrid Genetic Algorithms

A control system for automatically generating a train operation evaluation curve based on a hybrid genetic algorithm has been proposed in order to improve the safety of automatic train operation. The system is based on Internet of Things (IoT) mobile devices and utilizes various sensors such as accelerometers, gyroscopes, barometers, and GPS to collect real-time data on the driver’s acceleration, angular velocity, air pressure, and position, among other parameters. 5G wireless communication technology is used to achieve high-speed data transmission and real-time communication with the cloud. Based on cloud data, a spatial grid area planning model is constructed to automatically generate the train operation evaluation curve. Using a spatial dynamic programming method, the entire network of Electric Multiple Unit (EMU) trains is treated as a whole, and a dynamic model of the EMU train is constructed. The spatial area parameters of the EMU train’s automatic driving operation evaluation curve are combined with the dynamic model analysis method. By identifying and analyzing environmental parameters such as train speed and distance, the EMU train’s automatic driving operation evaluation curve is optimized. A hybrid genetic evolution learning optimization algorithm is used to fit the motion spatial parameters of the EMU train’s automatically generated driving operation evaluation curve, and a spatial behavior analysis simulation is created for the EMU train’s automatically generated control driving operation curve. Through the use of hybrid genetic evolution learning optimization technology, adaptive control and automatic driving operation curve simulation planning for the automatically generated driving operation evaluation curve of the maglev train are achieved, as well as simulation and algorithm optimization design for the automatically generated driving operation evaluation curve of the maglev train. The simulation results show that the method has good adaptability and strong automatic control capability. The experimental results demonstrate the control effect of the proposed method on energy consumption and train stopping error, indicating that the proposed method can effectively improve the parameter adjustment and offset correction ability of the evaluation curve generated by the high-speed train driving operation.

Rail Flaw B-Scan Image Analysis Using a Hierarchical Classification Model

AbstractAs railway traffic volumes and train speeds increase, rail maintenance is becoming more crucial to prevent catastrophic failures. This study aimed to develop an artificial intelligence (AI)-based solution for automatic rail flaw detection using ultrasound sensors to overcome the limitations of traditional inspection methods. Ultrasound sensors are well-suited for identifying structural abnormalities in rails. However, conventional inspection techniques like rail-walking are time-consuming and rely on human expertise, risking detection errors. To address this, a hierarchical classification model was proposed integrating ultrasound B-scan images and machine learning. It involved a two-stage approach—model A for fuzzy classification followed by Model EfficientNet-B7 was identified as the most effective architecture for both models through network comparisons. Experimental results demonstrated the model's ability to accurately detect rail flaws, achieving 88.56% accuracy. It could analyze a single ultrasound image sheet within 0.45 s. An AI-based solution using ultrasound sensors and hierarchical classification shows promise for automated, rapid, and reliable rail flaw detection to support safer railway infrastructure inspection and maintenance activities.

A multi modal fusion coal gangue recognition method based on IBWO-CNN-LSTM

Accurate identification of coal and gangue is a crucial guarantee for efficient and safe mining of top coal caving face. This article proposes a coal-gangue recognition method based on an improved beluga whale optimization algorithm (IBWO), convolutional neural network, and long short-term memory network (CNN-LSTM) multi-modal fusion model. First, the mutation and memory library mechanisms are introduced into the beluga whale optimization to explore the solution space fully, prevent falling into local optimum, and accelerate the convergence process. Subsequently, the image mapping of the audio signal and vibration signal is performed to extract Mel-Frequency Cepstral Coefficients (MFCC) features, generating rich sample data for CNN-LSTM. Then the multi-head attention mechanism is introduced into CNN-LSTM to speed up the training speed and improve the classification accuracy. Finally, the IBWO-CNN-LSTM coal-gangue recognition model is constructed by the optimal hyperparameter combination obtained by IBWO to realize the automatic recognition of coal-gangue. The benchmark function proves that IBWO is superior to other optimization algorithms. By building an experimental platform for the impact of coal and gangue falling on the tail beam of hydraulic support, multiple experimental data collection is carried out. The experimental results show that the proposed coal-gangue recognition model has better performance than other recognition models, and the accuracy rate reaches 95.238%. The multi-modal fusion strategy helps to improve the accuracy and robustness of coal-gangue recognition.

Research on the dynamic response of semi-enclosed sound barrier on viaduct under train pulsating wind and wheel-rail excitation

This study develops a coupled train-viaduct dynamic analysis model to investigate the dynamic response of semi-enclosed sound barriers on viaducts in light rail transit. During the iterative process of this model, fluctuating wind produced by the train is introduced and combined with wheel-rail loads to analyze the dynamic response of sound barriers. Based on the above model, in the time domain, there are significant differences in the vibration displacement and acceleration under combined excitation when the train head enters and the tail exits the sound barrier region. In the frequency domain, the vibration acceleration level under combined excitation increases significantly within the 0–5 Hz frequency. The parameter analysis results indicate that the total vibration acceleration level is higher at the measuring points close to the direction of travel. As the train speed increases, the effect of fluctuating wind becomes more significant in the low-frequency range (1–3.15 Hz). The smaller number of train formations also improved the vibration acceleration level difference between the combined and wheel-rail excitation.