Short video platforms have quickly emerged as a source of digital media consumption, which is overwhelmingly researched in terms of visual saturation, dependence on algorithms, and performance response. The auditory aspect, and, in particular, the homogenization of the background sound effects, is not researched thoroughly. This review will examine how user auditory fatigue can be caused by acoustic homogenization on short video platforms. It states that the repetition of similar background music structure, transition signal, loudness, timbral textural profiles, and emotionally coded sound patterns produces a platform-mediated auditory scenery with a lack of diversity and continued sensory demand. The article builds a conceptual model based on the insights of acoustics, psychoacoustics, media studies, human-computer interaction, and platform studies to construct a link between acoustic homogenization and auditory fatigue by proposing mechanisms, which include perceptual saturation, augmented listening effort, dynamic compression, enhanced temporal turnover, algorithmic reinforcement, and cross-modal overload. It also examines existing methods of measurement, such as the acoustic feature analysis, subjective fatigue evaluation, behavioral measures, and physiological measures, and points to the weaknesses of the existing empirical findings. It concludes the article by stating that auditory fatigue in short video use must be viewed as a cumulative and ecologically organized response, and not a one-sided response. This review is an interdisciplinary step forward by foreshadowing the aural ecology of short video platforms, as well as suggesting fatigue-conscious future research and platform design and governance of digital well-being.
This paper develops a mathematical optimization framework for broadband vibration attenuation of a simply supported thin plate using a sparse network of passive inerter-dashpot dynamic absorbers. The four attachment locations are fixed a priori away from nodal lines of the retained low-order modes; the optimizer allocates a total absorber mass limited to 4.2% of the plate mass and constrains tuning frequency to 75–580 Hz, damping ratio to 0.035–0.290, and inertance ratio to 0.05–4.50 over the selected 45–650 Hz numerical band. A modal Kirchhoff–Love model is condensed with frequency-dependent absorber impedances, and a nonsmooth Chebyshev H-infinity epigraph objective is treated by log-sum-exp smoothing, differential-evolution exploration, and projected local polishing. For the ten-mode benchmark, the robust design reduces nominal peak mobility from 91.889 dB to 85.096 dB: its 6.793 dB reduction exceeds modal tuning by 0.468 dB and the uniform-inerter layout by 1.604 dB. Across fifty material-and-damping perturbation scenarios, its 95th-percentile peak is 1.757 dB below modal tuning, indicating a practically relevant upper-tail benefit, although experimental validation remains necessary. Because the tenth retained plate mode is 385.356 Hz, results above that frequency are reported as exploratory pending a 15–20-mode convergence study.
Partially filled storage and transport tanks often radiate objectionable low-frequency sound because liquid sloshing, flexible shell modes, and uncertain damping interact in the same operating band. This paper develops a compact hydroelastic vibro-acoustic design method for a vertical cylindrical tank equipped with internal porous annular baffles and external viscoelastic rings. The fluid is represented by Bessel-based sloshing modes, the shell by a Ritz expansion, and the nonlinear free-surface contribution by a cubic Koopman–Galerkin lifting. Interval type-2 fuzzy sets describe epistemic uncertainty in fill height, fluid density, shell stiffness, loss factor, and baffle clogging. A deterministic design and a proposed fuzzy robust design are compared over alpha-plane uncertainty scenarios using peak sound power, band-averaged sound power, wall acceleration, and wave-height constraints. The proposed design reduces the 95th-percentile peak sound power from 57.84 dB in the uncontrolled tank to 48.55 dB, decreases the 95th-percentile band-averaged acoustic level from 28.47 dB to 26.73 dB, and lowers the 95th-percentile wall-acceleration level from 130.48 dB to 120.34 dB. Relative to the deterministic optimum, the robust design has a 0.216 dB higher nominal peak and a 0.173 dB higher 95th-percentile peak. Still, it lowers the 95th-percentile band mean by 1.059 dB and wall acceleration by 0.613 dB. The contribution is the integrated reduced-order design workflow; physical validation remains necessary.
Quasi-zero-stiffness (QZS) isolators possess a combination of high load capacity and low dynamic stiffness; however, their characteristics are sensitive to geometric nonlinearity, frequency-dependent dissipation, excitation amplitude and epistemic uncertainty. In this study, a new design scheme is proposed to produce the constant geometric restoring force of a three-spring QZS isolator subject to base excitation. Caputo fractional damping + exact first-harmonic stiffness (in terms of complete elliptic integrals) + multi-harmonic alternating frequency–time harmonic balance (to check the harmonic content and residual error). Interval type-2 fuzzy numbers account for uncertainty in the following quantities: geometric ratio, residual tangent stiffness, fractional damping intensity (or order), and base-motion amplitude. Differential evolution of upper-tail resonance risk and lower-tail isolation losses. A limiting-case validation against a literature benchmark verifies that the fractional-order models recover the classic viscously damped three-spring QZS equations for the case of approaching finite response amplitude and unity fractional order. For the 50 kg realization, the optimized design has a geometric ratio of 1.1824, residual stiffness of 0.02110, damping ratio of 0.02581 and fractional order of 0.9063. Its 95th-percentile peak transmissibility is 3.358 across 120 outer-footprint scenarios, which is 28.9% below the optimized viscous-limit QZS reference. Adverse-tail isolation onset is 23.3% earlier than deterministic fractional tuning completion, and adverse-tail mean attenuation improvements are as high as 1.08 dB. The five-harmonic verification ensures that the ratio of the third harmonic stays below 0.31% in all ranges mentioned in operating range.
This study presents a targeted-band design method for suppressing flexible-rotor vibration using a fractional-order fuzzy tuned mass damper. A force-excited rotor-support mode represents the host machine, while the absorber branch comprises a spring and fractional dashpot. The absorber parameters are selected to minimize the primary displacement peak while remaining robust to bounded uncertainty in mass ratio, tuning ratio, damping index, fractional order, support stiffness, and force level. Triangular fuzzy numbers are propagated through alpha-cut intervals, and the resulting frequency-response envelopes are optimized by a weighted objective that penalizes high peaks and wide uncertainty spreads. Closed-form frequency-response relations evaluated directly at interval corners keep the procedure transparent and reproducible. For a rotor-support oscillator with natural frequency 9.762 Hz, primary mass 42 kg, primary stiffness 158,000 N/m, force amplitude 58 N, and mass ratio μ = 0.05, the optimized design yields a central peak displacement of 1.929 mm and an alpha-zero upper peak of 3.154 mm, compared with 10.197 mm for the uncontrolled system and 2.028 mm for the deterministic Den Hartog reference. Evaluation at μ = 0.03, 0.05, and 0.07 shows consistent peak reduction and useful trends for initial parameter tuning. The method therefore provides bounded vibration envelopes, tuning guidance, stroke estimates, and sensitivity rankings for uncertain rotating-machine data within the specified target band.
Industrial rotating machinery plays a pivotal role in global energy infrastructure, yet conventional vibration monitoring systems often operate as black boxes, providing limited interpretability and failing to leverage the rich multi-sensor data available in modern plants. This paper introduces a novel framework that integrates multimodal sensor fusion—combining accelerometer, acoustic, thermal, and operational data—with explainable artificial intelligence (XAI) and multi-criteria decision analysis. The core engine is a domain-collaborative multimodal transformer that jointly processes heterogeneous time-series and image-based streams, producing fault classifications alongside SHapley Additive exPlanations (SHAP)-based feature attributions and natural-language diagnostic narratives. The framework is validated on a 250 MW combined-cycle gas turbine power plant with 24 months of operational data. Experimental results demonstrate a fault detection accuracy of 94.2%, a 14.5% improvement over vibration-only baselines, while achieving the highest interpretability score (5/5) among compared methods. Decision Making Trial and Evaluation Laboratory (DEMATEL) causal analysis identifies diagnostic transparency and system reliability as primary drivers of regulatory compliance. The primary contribution is an open-source, scalable blueprint for trustworthy AI in industrial vibration monitoring, addressing the urgent need for transparent, auditable, and human-centered decision support in critical energy assets. The proposed framework achieves a balanced integration of three critical dimensions: diagnostic accuracy and interpretability, technical performance and regulatory compliance, and automated inference and human oversight. Based on these findings, we recommend that industrial operators for finance risk optimization: (1) deploy multimodal sensor arrays combining vibration, acoustic, thermal, and operational sensors; (2) implement explainable AI protocols utilizing SHAP-based feature attribution; and (3) adopt DEMATEL-derived priorities for risk-informed maintenance scheduling.
The rapid evaluation of interior aerodynamic noise during the Concept A Surface design stage is important for vehicle acoustic development, but conventional methods are limited by high cost and low efficiency. This study proposes a convolutional neural network transformer-based prediction method for vehicle interior wind noise by integrating vehicle styling features and acoustic technical parameters. An optimal Latin hypercube sampling method was used to generate design combinations, and wind tunnel tests were conducted at 120 km/h. Key vehicle styling parameters, including A-pillar geometry, side mirror dimensions, front windscreen angle, side mirror-to-body spacing, and side window inclination, together with glazing material properties, glass thickness, acoustic transfer function, and interior reverberation time, were selected as input features to predict the driver’s left-ear wind noise spectrum. Based on five-fold cross-validation, the proposed model was compared with convolutional neural network (CNN), long short-term memory (LSTM), and transformer models. The CNN-Transformer model achieved the best performance, with mean absolute percentage error (MAPE) and root mean square error (RMSE) values of 2.23% and 0.94 dB, respectively. Compared with the transformer, CNN, and LSTM models, the proposed method reduced MAPE by 24.91%, 36.29%, and 49.59%, and reduced RMSE by 22.95%, 35.17%, and 48.07%, respectively. The model also maintained reliable performance on an independent test set, with MAPE and RMSE values of 4.82% and 1.44 dB. The mean impact value method was further applied to identify the influence of design parameters on interior wind noise, guiding for early vehicle acoustic optimization.
Electrostatic precipitators (ESPs) are significant devices for particulate control in power generation, cement production, steel manufacturing, and other high-emission sectors. Future ESP performance should be evaluated not only on particle collection efficiency but also on electrical energy consumption, vibration stability, noise levels, reliability, and adaptive functionality. We proposed a three-layered framework for perception, analysis, and execution in intelligent ESPs. The perception layer combines electrical, emission, process, vibration, acoustic, and maintenance data; the analysis layer applies signal processing, artificial intelligence, digital twins, and multi-objective optimization; and the execution layer provides adaptive voltage control, rapping optimization, fan-speed control, vibration mitigation, active noise control, and predictive maintenance. The quantitative data show that intelligent electrical optimization can reduce ESP energy consumption by 35.50% and improve emission compliance from 95% to 100%. Approximately 43% energy saving was achieved by deep-learning-assisted voltage optimization in a 330 MW coal-fired power plant ESP. FFT, wavelet transform, RMS tracking, CNNs, LSTM models, and autoencoders can be used as diagnostic methods to detect imbalance, resonance, bearing faults, and fan irregularities, thereby reducing vibrations. Hybrid passive-active control for noise reduction can attenuate low-frequency duct noise by more than 10 dB and high-frequency components by more than 20 dB. This review highlights the absence of field-validated collaborative optimization as a significant knowledge gap and suggests digital twins, edge computing, 5G/6G communication, and multi-objective control as promising avenues for future research.
To address the issues of low fitting accuracy, parameter selection relying on empirical judgment, and difficulties in quantifying model robustness in the traditional Sadovsky blasting vibration prediction formula, this study proposes a method for modifying the peak particle velocity prediction model that balances fitting capability and robustness by integrating Bayesian theory with dimensional analysis. A model prior distribution incorporating multiple on-site blasting parameters is constructed using the dimensional pi theorem. Within the Bayesian framework, the maximum likelihood estimation, Occam factor, and posterior credibility of the model are calculated to achieve automatic selection of influencing factors and optimization of the model structure. Based on 88 sets of measured data from an open-pit quarry, with 70 sets used as training samples and 18 sets as validation samples, model training and validation are conducted. The results show that the coefficient of determination R2 of the Bayesian modified model increases from 0.7749 obtained by the traditional Sadovsky formula to 0.8576. The Occam factor can effectively characterize the robustness of the model. The preferred model "1 2 4" incorporates empirical formulas for correcting the resistance line, spacing between rows, and borehole diameter. This model achieves an optimal balance between prediction accuracy and robustness, and its prediction stability is significantly superior to that of traditional empirical formulas. This method provides a theoretical basis and engineering reference for accurate prediction and safety control of blasting vibrations.
The article analyzes the characteristic features of the vibration cutting process with ultrasonic frequency (UF), which is one of the most effective methods that reduces labor costs and improves their quality and reliability during the mechanical processing of particularly hard and brittle materials. To stabilize the amplitude and frequency, changing as a result of changing the resonant frequency of the vibrator-concentrator-part (VCP) oscillatory system during processing, linear and relay-type generators are proposed that operate on the basis of excitation of the natural frequency of the mechanical vibration system VCP, and these generators are modeled and synthesized using MATLAB/Simulink. With the development of modern machinery and technology, new materials with improved mechanical properties are created, on the basis of which critical parts are manufactured, where high accuracy, quality, and reliability indicators are required. In the machining process, to stabilize the amplitude and frequency, which change as a result of the change in the resonant frequency of the vibrator-concentrator-part (VCP) oscillatory system, generators of the linear and relay type are proposed, which work on the basis of the excitation of the own frequency of the mechanical vibration system of the VCP. These generators are modeled and synthesized in MATLAB/Simulink. The obtained mathematical models and curves of the transition process show that, despite the change in the disturbing factor when the coefficient of friction is changed, conservatism is ensured, as a result of which there is an opportunity to improve the quality of surfaces machining of parts.
Heart-sound recordings are highly susceptible to environmental and physiological noise, which complicates clinical interpretation and reduces the reliability of automated diagnostic systems. Effective denoising is therefore essential for preserving waveform morphology and enabling accurate feature extraction. This study proposes a heart-sound-specific wavelet approach and evaluates its denoising performance in comparison with conventional wavelets and support vector machine (SVM)-based methods. The method was assessed using publicly available datasets, including PASCAL and PhysioNet, which provide diverse normal and pathological phonocardiogram (PCG) recordings. Uniform and Gaussian white noise were added at varying signal-to-noise ratios (SNRs) to simulate realistic acquisition environments. Denoising performance was quantified using cross-correlation coefficients, SNR improvement, root-mean-squared error (RMSE), and mean absolute error (MAE). Results demonstrate that the proposed heart-sound wavelet achieved superior noise-suppression capability and a 7% performance gain over commonly used Db and Bior wavelets, while maintaining waveform integrity. Subsequent classification experiments showed that denoising quality directly influenced diagnostic performance: the model achieved 0.87 accuracy, 0.81 precision, and 0.83 sensitivity on the PASCAL dataset, and 0.997 accuracy, 0.946 sensitivity, and 0.944 precision on PhysioNet. These findings highlight the potential of tailored wavelet-based denoising to enhance automated heart-sound analysis and support more robust clinical and embedded diagnostic applications.
In recent decades, the rapid development of transportation infrastructure safety, such as highways, bridges, and tunnels has greatly promoted the development of the regional economy. The structures with safety hazards and emergencies need continuous monitoring over time. The integrated artificial intelligence algorithms with sensor responses can provide real-time information for further analysis and decision-making for the transportation system, improving the circulation efficiency of the transportation network, ensuring the stability of road structures, and avoiding irreparable damage. This paper aims to develop an efficient and low-cost method to help detect early-stage transportation infrastructure damage through permanent or periodic monitoring. In this research, we used LiDAR scanning units (terrestrial LiDAR fixed on holders and movable units fixed on UAVs) integrated with a novel deep neural network (DNN) for structural monitoring of bridges based on the 3D mapping of bridge displacement compiled from LiDAR scanning over time. The monitoring model is based on a recurrent neural network with long short-term memory blocks (RNN-LSTM) since the LiDAR scanning datasets have a time-dependent and memory-dependent behavior. The response of the proposed DNN achieved a high accuracy rate, regression rate, and F-score equal to 96.43%, 93.77%, and 91.65%, respectively. A deep analysis of the confusion matrix and a side-by-side look at predicted and actual conditions highlight how well the model can tell apart different traditional methods to estimate the bridge displacement in literature. So, the data from LiDAR and DNN models can be combined to analyze the monitoring of transportation infrastructure.
The presence of stationary or non-stationary noise substantially impairs speech intelligibility and its perceptual quality in present-day communication systems. Many existing spectral-domain methods like Wiener filtering and spectral subtraction rely on strong statistical assumptions about complex-valued data, while state-of-the-art neural models such as Conv-TasNet, DCCRN, and MetricGAN+ achieve remarkable performance gains for speech enhancement, but tend to suffer from heavy computational complexity and inference latency due to their complex architecture. In this paper, we present a new supervised non-causal WaveNet with parallel target-field prediction for efficient end-to-end speech denoising. The proposed model captures both past and future temporal context by enabling symmetric receptive fields and removing autoregressive dependencies. Traditional autoregressive WaveNet models generate samples one at a time, which require considerable redundant convolution operations due to the repetitive nature of their structures and tasks, while the presented method is capable of predicting a target field of samples in a single forward pass, allowing for parallel inference. We validate the proposed model on the NSDTSEA dataset for input SNR levels of 2.5 dB to 17.5 dB and under stationary as well as non-stationary noise types. We showcase experimental results that prove that the proposed framework consistently outperforms strong classical baselines, with over 1.22 dB gain in output SNR and 46% reduction in Mel Cepstral Distortion (MCD) under non-stationary noise. Moreover, when compared to state-of-the-art deep learning models such as Conv-TasNet, DCCRN, and MetricGAN+, the proposed method achieves a viable trade-off between performance, robustness, and computational efficiency.
The physical work environment significantly influences worker performance, with workplace acoustics playing an important role in cognitive functioning and productivity. This study investigated the effect of music genre as an engineered acoustic intervention on productivity during repetitive industrial tasks. Unlike previous studies emphasizing psychological outcomes, this research integrates occupational acoustics, industrial ergonomics, and productivity engineering to evaluate music as a controllable workplace sound variable. A repeated-measures experimental design was conducted under four auditory conditions: no music, jazz, pop, and instrumental music. Productivity was assessed using output and work cycle time during repetitive manual tasks. Data were analyzed using one-way repeated-measures analysis of variance (RM-ANOVA) followed by Bonferroni-adjusted pairwise comparisons. Mauchly's test confirmed that the sphericity assumption was satisfied (W = 0.964, p = 0.119). Music genre had a significant effect on productivity (F(3,897) = 412.68, p < 0.001, partial η² = 0.58). Pop music produced the greatest productivity improvement (33.82%), followed by jazz (13.03%) and instrumental music (10.45%). Bonferroni comparisons showed that pop music significantly outperformed all other auditory conditions, whereas instrumental music did not differ significantly from the control condition. These findings demonstrate that appropriately designed auditory environments can enhance productivity during repetitive work. The study contributes to occupational acoustics by positioning music as an engineered environmental variable that supports human-centered industrial design and operational performance.
To tackle the intractable problems including weak fault feature extraction and evolution uncertainty quantification for complex systems in strong noise environments, a novel method for weak fault diagnosis and evolution analysis is proposed. This method integrates the fuzzy structured element (FSE), cloud model (CM), and Space Fault Network (SFN). The method centers on adaptive wavelet denoising, fault feature cloudification, and SFN probability propagation. The fault signal under strong noise is reconstructed by optimizing the wavelet threshold with the FSE. The uncertainty encapsulation of the peak factor of fault features is realized based on the CM to establish the feature CM. The fault event topology is constructed relying on the SFN. The quantitative transfer of uncertainty in the fault evolution is achieved combined with cloud algebra. Verified by the inner ring pitting fault of axle box bearings, the results demonstrate that the proposed method can extract the fault characteristic frequency of 250.5 Hz. The derived fault probability CM (0.680, 0.059, 0.023) accurately quantifies the system risk level. This result is consistent with the actual fault evolution law in engineering practice. This method provides technical support for early fault warning and maintenance of complex industrial system. Furthermore, comparative experiments confirm its superiority over traditional methods in noise suppression and feature retention. Parameter analysis is also discussed to improve engineering generalization.
As wind turbines increasingly participate in Primary Frequency Regulation (PFR) to support grid frequency stability, the accompanying mechanical load variations pose a potential threat to structural reliability. However, the underlying aero-electro-mechanical coupling mechanisms by which PFR-induced power and torque fluctuations affect tower dynamics have not been fully clarified. This study develops a multidimensional analytical model of a wind turbine with PFR to reveal the transmission path from grid frequency deviations to generator torque variations and subsequently to tower side-to-side (SS) bending moments. Frequency-domain and time-domain analyses show that generator torque, rather than aerodynamic thrust, is the dominant excitation source for tower SS vibration during PFR. The results further indicate that this excitation is highly phase-sensitive: the amplification or suppression of tower vibration depends on the instantaneous phase alignment between the torque disturbance and the tower’s natural sway cycle. Therefore, identical PFR commands may lead to substantially different structural responses when activated at different instants. The analytical conclusions are validated using high-fidelity FAST co-simulations under stochastic wind and grid-disturbance scenarios. Quantitative fatigue evaluation shows that uncoordinated PFR may increase the tower SS Damage Equivalent Load (DEL) by up to 600% in low-wind-speed regimes. These findings demonstrate that future wind turbine PFR controllers should incorporate phase-aware coordination strategies to reduce structural resonance and fatigue risks while maintaining effective grid frequency support.
Honeycomb structures have been widely used in many industrial fields due to their excellent properties. However, it is always challenging to rapidly and accurately detect defects such as debonding in the structure, especially in the in-service situation. In response to this, we have proposed an acoustic testing method without coupling agents based on the acoustic band gap feature in the structure. However, the influence of test parameters such as signal excitation and reception on the band gap feature has not been comprehensively and thoroughly investigated, and the parameters have not been optimized. In this paper, the transmission frequency response (TFR) curves were measured, and the band gap features were investigated at different parameters. The results demonstrated that the band gap feature changes a little with the pressure between the probes and the specimen, the amplitude of the exciting signal, the sweep duration, and the detection direction. While it changes significantly with the exciting-receiving distance. Experimental results demonstrated that to form a stable band gap feature, the wave should propagate through at least three honeycomb unit widths before being received. Further analysis indicates that the defect resolution of the proposed method is about two honeycomb unit widths. This work can be used to select the proper detection parameters and further improve the reliability and efficiency of this technique.
This study presents an integrated multiphysics framework for the co-design of flow field architectures in redox flow batteries, aiming to simultaneously optimize electrochemical performance and structural acoustic reliability. A high-fidelity numerical methodology is developed to couple electrolyte hydrodynamics, species transport, electrochemical kinetics, and flow-induced structural vibration, enabling a comprehensive assessment of both electrochemical and mechanical behavior. To overcome the trade-off between mass transfer uniformity and flow-induced noise, the framework employs biomimetically inspired channel topologies—derived from natural fluid transport systems—that enhance homogeneity of species distribution while inherently suppressing flow-induced vibration and acoustic emissions. As a result, both electrochemical polarization and mechanical excitation are reduced. The proposed designs are validated through comparative computational fluid dynamics (CFD) and structural dynamics analyses. Results demonstrate measurable gains in round-trip efficiency, reduced pumping losses, and improved operational stability compared to conventional flow field configurations. By bridging biomimetic fluidic design with rigorous dynamic structural analysis, this work provides an analytically sound and experimentally viable pathway toward next-generation grid-scale energy storage systems that are not only efficient and durable but also operate quietly with reduced mechanical fatigue. The framework thus addresses critical barriers to the widespread deployment of redox flow batteries, particularly in noise-sensitive or vibration-prone environments.
This paper presents the theoretical foundations for the nonlinear modelling of liquid sloshing dynamics in shells of revolution subjected to combined horizontal and vertical excitations. The proposed mathematical model integrates a spectral approach, the boundary element method (BEM), and a modal representation of the solution in generalized coordinates. This unified framework enables consistent analysis of both linear and nonlinear sloshing phenomena. The spectral boundary-value problem is reduced to a system of singular integral equations defined on the free and wetted surfaces of the shell. The 2π-periodicity of the integral operators is rigorously established, and the structure of the kernel singularities is identified. This ensures mathematical consistency and supports efficient numerical implementation. Natural sloshing modes and frequencies are computed using a BEM-based solver. Based on these eigenmodes, a modal expansion is constructed, leading to reduced-order systems of nonlinear ordinary differential equations. Governing equations are derived for both linear and nonlinear formulations, allowing detailed investigation of sloshing responses under simultaneous horizontal and vertical excitations. Rayleigh damping is incorporated into the modal system; its applicability and physical justification are discussed, along with its limitations in capturing energy dissipation in violent sloshing processes. Numerical results for rigid cylindrical shells illustrate the significant influence of nonlinear modal interactions and combined loading on free-surface elevation. The developed approach provides an effective tool for predicting complex sloshing behaviour beyond the capabilities of purely linear theory.
Railway vehicles operate under continuously changing wheel-rail contact, track geometry, loading, speed, and environmental conditions, making vibration behavior a direct driver of ride comfort, running safety, structural fatigue, and maintenance demand. Existing studies often examine vehicle vibration, structural reliability, and dynamic performance separately, which can obscure the causal path from track excitation to vehicle response, fatigue damage, performance degradation, and maintenance action. This review synthesizes these topics within an integrated life-cycle framework. It first examines major vibration excitation mechanisms, including track irregularities, wheel and rail defects, turnouts, transition zones, traction and braking forces, and aerodynamic disturbances. It then reviews modeling and analysis methods, including multibody dynamics, finite element analysis, vehicle–track coupled models, co-simulation, monitoring-based validation, and uncertainty analysis, with emphasis on their ability to connect dynamic loads with stress histories and fatigue reliability. Structural reliability is discussed as the mechanism by which vibration-induced loads become service risk, while dynamic performance is evaluated through ride comfort, running safety, hunting stability, vibration control, and multi-objective optimization. Furthermore, key limitations such as simplified contact assumptions, insufficient full-scale validation, weak coupling between monitoring data and fatigue models, limited interpretability of AI diagnosis and immature digital-twin implementation are identified in this review. Finally, future directions are proposed toward physics-informed modeling, uncertainty-aware reliability assessment, intelligent monitoring, adaptive vibration control and reliability-centered maintenance for safer, more durable, and sustainable railway vehicles.
