Aiming at the problems of failure correlation, few fault data samples, and too simple evaluation indicators in the existing Failure Mode, Effects and Criticality Analysis (FMECA) of CNC machine tools, a comprehensive risk dynamic assessment method for CNC machine tools is proposed that integrates the degree of fault impact,failure frequency and consequence evolution characteristics. DEMATEL/ISM/ANP, Bayes, Copula, statistical analysis, cloud model and other methods are used to evaluate the failure impact degree, failure frequency,failure rate, failure maintenance time,failure maintenance difficulty,maintenance ergonomics requirements and maintenance cost. Accordingly, the data envelopment analysis method is introduced to form an index evaluation matrix and evaluate the failure risk. Finally, an example of a series of domestic machining centers is used to verify the effectiveness of the method.
The trend towards zero-emission vehicle propulsion results in a fuel cell system working in conjunction with an electric motor and battery system. The research carried out in this study concerns the evaluation of fuel cell performance made without any interference to the drivetrain. The article aims to fill a gap in research on the efficiency of fuel cells vehicle using a non-invasive method in real traffic conditions, which is lacking in the literature. The research was conducted using a 128 kW FCHEV under typical traffic conditions. As part of the work, the propulsion efficiency was determined for each driving phase. The following fuel cell efficiency values were obtained in the urban, suburban and highway phases: 51, 56 and 66% respectively. The values obtained are very close to the direct tests, while completely eliminating the need to modify the powertrain in order to determine the fuel cell efficiency.
Overhead Transmission Lines (OTLs) are chronically subjected to fatigue damage induced by Aeolian vibration and extreme wind-ice loads. However, existing design codes and reliability assessment methods often overlook the coupling effects of these two factors. This paper proposes a reliability framework integrating regional climate and full life-cycle effects, applied to a Transmission Tower-Line System (TTLS) under combined wind and ice loads. Based on meteorological data from Shenyang and Xianning (1975–2025), a joint probability model of wind speed and direction is constructed using the Frank Copula to quantify Aeolian vibration probability. Nonlinear dynamic time-history simulations are employed to analyze conductor strength degradation. Integrating fragility estimates and reliability indices reveals the evolution of structural safety margins over the service life. Results show that fatigue damage significantly reduces OTL load-bearing capacity. As service extends, loads sustainable at the initial design stage evolve into critical loads, precipitating catastrophic failure.
Traditional reliability models assume subsystem failures are independent, ignoring dependencies and leading to bias. To address this, a dependent reliability modeling method based on Copula and multi-index fusion is proposed. The rank-sum method integrates multi-dimensional metrics to determine optimal lifetime distributions for each subsystem. The maximum information coefficient quantifies fault dependency between subsystems and identifies dependent pairs. The Archimedean Copula models the dependence structure, parameters estimated via maximum likelihood, and optimal Copula selected using goodness-of-fit metrics. Empirical studies with real failure data show that, compared to independent models, the proposed method improves assessment accuracy throughout the period, confirming dependency modeling provides a positive effect. Based on actual data, a framework covering lifetime fitting, dependency identification, and optimal Copula selection is constructed, offering a practical tool for systems with fault dependencies and reference for maintenance and risk control.
To overcome the prolonged test durations and large sample sizes inherent in conventional accelerated degradation testing (ADT), this paper develops a novel reliability assessment framework based on Weibull pseudo-failure lifetimes. Uniquely, the proposed method enables high-confidence reliability assessment using data from only two accelerated stress levels. First, a mathematical relationship between the use-level reliable life and the Weibull distributional parameters at accelerated levels is established. Subsequently, a joint confidence distribution for these parameters is constructed, facilitating an explicit formula for the exact one-sided lower confidence limit of the reliable life. Monte Carlo simulations and case studies demonstrate that the proposed framework significantly reduces the sample size and duration, while enhancing theoretical rigor and assessment precision compared to traditional multi-level asymptotic methods.
Accurate monitoring of industrial equipment is important for production safety and reliability. This paper proposes a physical-causal guided adaptive time-frequency and hypergraph co-evolutionary modeling method. The method constructs neural network representations with parameterized basis functions, enabling adaptive time-frequency decomposition with energy conservation and modal orthogonality constraints. Differentiable physical projection operators ensure hard constraint satisfaction during training. A transfer entropy-based causal discovery mechanism converts causal relationships into hyperedge generation rules for dynamic hypergraph construction. Bidirectional cross-attention mechanisms achieve collaborative updates between time-frequency and hypergraph features. Experiments achieve 98.73% fault diagnosis accuracy on CWRU dataset, 91.77% cross-working-condition transfer accuracy, 82.56% accuracy at -6dB SNR, and 82.4% cross-dataset generalization accuracy on unseen equipment types.
Predictive maintenance improves reliability and reduces downtime in modern manufacturing systems. However, many studies rely on laboratory datasets or single-component monitoring, limiting their applicability to complex industrial environments. This study proposes a predictive maintenance framework for a multi-pass wire drawing machine using vibration and motor current signals from a real industrial production line. A Composite Health Index (CHI) is developed to transform multi-motor sensor data into an interpretable machine-level degradation indicator. The extracted features are used to train ensemble machine learning models including Random Forest, Extra Trees, and XGBoost, whose outputs are combined through a weighted hybrid ensemble model. A decision-layer mechanism with smoothing and temporal filtering is applied to reduce false alarms while preserving detection capability. Experimental results show that the model achieves a recall of 0.90 and an F1-score of 0.75, demonstrating its effectiveness for industrial applications.
Addressing the critical challenge of low-cycle fatigue (LCF) in electrohydraulic hammers induced by repetitive sub-critical pressure fluctuations (35-40 MPa), which remains unmitigated by conventional relief valves due to an inherent response "dead zone", this study proposes a novel dual-stage relief valve. The core innovation resides in a fixed geometric differential area (Delta A) designed to decouple high-and low-pressure regulation mechanisms. This architecture synergizes a direct-acting stage for catastrophic surges (calibrated >40 MPa) with a pilot-operated differential stage for sub-critical transients (35-40 MPa), thereby eliminating the regulation blind spot. The structural parameter Delta A is rigorously optimized via theoretical derivation and sensitivity analysis to ensure stable responsiveness across the target pressure range. Validated through AMESim/Simulink co-simulation and high-precision experiments, the valve achieves a rapid response time of 20ms, representing a 40% improvement over conventional valves, and reduces pressure fluctuation amplitudes from 5 MPa to 1.5 MPa. Furthermore, stress analysis and modified Miner's rule predictions demonstrate that suppressing these fluctuations reduces cyclic stress from 108.1 MPa to 43.2 MPa. Consequently, the predicted service life of critical components is extended from 9.6 & times;10(4) cycles to 9.8 & times; 10(6) cycles, an increase of nearly two orders of magnitude. This design establishes a paradigm shift from reactive overload protection to proactive stability management, significantly enhancing the reliability of heavy-duty hydraulic systems.
To address the issue that the classical LSE and MLE neglect global errors when estimating the two parameters of the log-normal distribution with small samples, a three-parameter estimation method integrating empirical distribution function model optimization, threshold parameter estimation, and cumulative sum of squared errors minimization is proposed. The initial dataset is derived via inverse transformation of the empirical distribution, and the empirical distribution model is optimized using the linear correlation coefficient. The interpolation method is employed to estimate the threshold parameter corresponding to the maximum linear correlation coefficient. The particle swarm optimization （PSO） algorithm optimizes the location and scale parameters to minimize the cumulative sum of squared errors. The K-S test is applied to evaluate the fitting performance, and the RMSE is used as an indicator to verify the accuracy of the proposed method. An empirical study is conducted by combining 13 sets of actual lifetime data and simulated data of a certain automotive battery.
In recent years, the problem of vandalized graffiti damaging walls, facades, and railway carriages has been growing worldwide. This problem primarily affects users of historic buildings and railways. The article addresses the problems of removing graffiti from railway carriages, which has a significant impact on their exploitation and additional costs. To preserve or limit possible vandalism, special anti-graffiti coatings are used. Several types of coatings are typically available on the market for specific practical applications. This article presents an original, quantitative, easy-to-use, and universal method for determining the best coverage among those considered, taking into account the user-defined essential characteristics that the coating should meet and their importance (weight). The following parameters were chosen: adhesion, nanohardness, hardness, total surface energy, erosion resistance, corrosion rate, and material price. The method's application was demonstrated by comparing four anti-graffiti coatings from two manufacturers.
The induction motor serves as the core power component in new energy vehicle drive systems. Any malfunction of the drive motor will directly undermine the operational reliability of the vehicle, potentially leading to drive system failures or, in severe cases, significant safety hazards endangering the lives of drivers and passengers. To address the issue of faint fault signatures within motor current signals, an image-based fault characterization method synergized empirical mode decomposition (EMD) with modified symmetrized dot pattern (MSDP) was proposed. Subsequently, a dedicated convolutional neural network (CNN) architecture with automated image feature extraction capabilities was engineered to classify motor health conditions, including healthy operation, bearing fault, and rotor broken bar. Finally, the efficiency and precision of the EMD-MSDP-CNN model for induction motor fault diagnosis were validated through experimental data, demonstrating a robust average diagnostic accuracy of 93.21% across diverse fault scenarios under multiple operating conditions.
The transition from Time-Based Maintenance (TBM) to Condition-Based Maintenance (CBM) remains a challenge for the railway sector due to stringent safety regulations. This paper addresses this gap by presenting a framework for diagnostic monitoring of traction motor cooling fans in Electric Multiple Units. Using vibration analysis and adaptive statistical thresholds, the study demonstrates increased sensitivity compared to ISO 10816-3 limits. Research on 42 fans revealed that the proposed method achieved a 92.9% agreement with normative classifications while identifying early-stage degradations. A crucial part of the study is the formal analysis of change significance in accordance with the 402/2013 (CSM RA) regulation, providing a systematic approach to managing safety risks during technology implementation. The article provides a practical roadmap for integrating this 19-minute procedure into routine light maintenance, offering a scalable solution to overcome formal and technical barriers in adopting predictive maintenance for safety-critical systems.
Accurate remaining useful life (RUL) prediction of AC contactors is essential for efficient operation and maintenance of the manufacturing system. Existing methods cannot adequately capture the degradation of AC contactors due to their inability to depict the special characteristic of zero and bounded arcing Joule integrals (i.e., degradation increments). To tackle this problem, the physical model of arcing Joule integrals is first derived through arcing mechanism analysis. Bounds of arcing Joule integrals are obtained by introducing critical breaking phase angles to the physical model, and four arcing modes are identified. A method for measuring the similarity between arcing modes is proposed, and then an arcing mode similarity based discrete-time Markov chain is constructed to depict arcing mode transitions. Motivated by zero and bounded arcing Joule integrals, an increment process with zero and bounded increments is proposed to characterize the degradation of a single-phase contact pair. Finally, the superiority of the proposed method is illustrated by real and numerical cases.
In dynamic confrontation scenarios, units within combat system of systems (CSoS) exhibit dual states: mission-active and standby. Standby units, serving as critical reserves for sustained combat operations, significantly influence system resilience continuity. Existing dynamic resilience research primarily assesses system capability through mission-active units' performance, neglecting the quantification of standby units' contributions to resilience evolution. To bridge this gap, this study proposes a dual-dimensional resilience assessment framework that integrates real-time mission performance metrics with the inherent reconnaissance and strike capabilities of standby units. This approach simultaneously quantifies instantaneous combat capability and latent recovery potential. Additionally, addressing the sensitivity degradation in resilience assessment due to cumulative disruption effects in prolonged confrontations, we design a dynamic phase-segmented resilience evaluation model (DPS-RE) that incorporates dynamic, time-varying performance baselines. Simulation experiments verify the regulatory influence of standby units on CSoS degradation and validate the effectiveness of targeted resource replenishment strategies. Furthermore, utilizing multi-domain joint combat simulation data, the DPS-RE model effectively identifies vulnerability windows and provides early warnings for resource supplementation, offering robust decision support in dynamic adversarial environments.
Helicopter pilots operate in a vibratory environment that substantially exceeds that of fixed-wing aircraft, creating stringent demands for the durability and dynamic stability of head-supported equipment. This study presents an experimental investigation of the vibration behaviour of prototype helicopter pilot goggles tested on an electrodynamic exciter over a frequency range exceeding applicable defence standards. The aim was to verify assembly integrity, assess the short-term strength of load-bearing components, identify natural frequencies of the goggle system, and generate reference data for future finite-element model validation. Sinusoidal excitations up to 20 g were applied to a steel mounting plate, a dedicated holder, and the complete goggle assembly in two extreme optical-track configurations. Acceleration and displacement responses were recorded using Bruel & Kj ae r accelerometers and processed through custom software to obtain amplitude-frequency characteristics and resonance peaks. The results indicate significant amplification of vibration in selected structural regions, particularly within the 50-250 Hz and 450-500 Hz bands, corresponding to natural modes of both the holder and the goggles. These findings confirm the sensitivity of the system to vibratory in-puts comparable to those encountered in helicopter cockpits and provide a validated experimental basis for subsequent high-fidelity numerical modelling and design optimisation.
This article presents an analysis of vibration measurements of a ship's Voith Schneider Propeller (VSP) system across the full range of the propulsion engine's load. Determining the vibration parameters serves as a form of operational diagnostics for the structural components of this modern propulsion system. This form of diagnostics, based on the analysis of vibration accelerations in three directions relative to the ship's axis as a carrier of information about potential malfunctions and damage, has not yet been given much attention in the literature. The limited number of descriptions and the lack of established vibration signatures for specific VSP malfunctions prompted an examination of the capabilities and benefits of using vibroacoustic methods to analyze the technical condition of a modern vessel's propulsion system. As part of the research, vibration signals were measured and recorded during sea trials on two propulsion units simultaneously across all possible load ranges. Analysis of selected parameters revealed that the propulsion system under examination experiences load levels that, for the propellers in question, should be kept as brief as possible due to their destructive effects, manifested by exceeding permissible acceleration values. The recorded and processed acceleration signals in three directions X (L), Y (H) and Z (V) provide a picture of the exceedances of the permissible operating levels described in standards and the manufacturer's documentation. Particular attention should be paid to the vibration velocity level in the measurement axis, the Z-direction (V), which shows the greatest variation compared to the other measurement directions. In this measurement direction, at a minimum propeller speed of 32 rpm, play in the drive system can be detected (presence of amplitude peaks), and at a maximum propeller speed of 99 rpm, the technical condition of the propeller can be assessed.
Fault diagnosis of in-wheel motor bearings is challenging due to weak fault features and non-stationary vibration signals under complex operating conditions. To address the limitations of conventional models in transient impact modeling and feature representation, this study proposes an enhanced Swin Transformer-based fault diagnosis framework. The proposed method integrates a multi-scale convolutional feature-enhanced feed-forward network (MSCF-EFFN) to improve shallow cross-scale representation, a unified deformable shifted window multi-head self-attention (DSW-MSA) framework to adaptively capture irregular transient impact features, and a depth-wise convolution attention module (DWCAM) to refine deep feature selection. The model is validated on a self-built dynamic test bench covering nine bearing health states and 28 operating conditions, achieving an average accuracy of 98.7% and a peak accuracy of 99.12%. Comparative and ablation studies demonstrate superior accuracy, robustness, and convergence performance over existing models.
This study presents the application of the Bayesian method to estimate the probability of critical marginal leakage in dental restorations. Aim of the study was to evaluate the effectiveness of the Bayesian method in estimating the probability of leakage associated with gap in the subsurface layers of composite restorations, based on observations of their surfaces. The study, involving the construction of a Bayesian network model and probabilistic inference, was based on the results of experimental research. These results were used to verify the a priori and a posteriori probabilities of damage to the tooth-restoration system in three zones of the tooth’s anatomical structure and across four load level intervals; using a Bayesian classifier, the study demonstrated the potential for diagnosing critical subsurface leaks. The results obtained show that surface fissure parameters can support the probabilistic diagnosis of leakage throughout the entire volume of the restorative filling, thereby assisting in clinical decision-making regarding the replacement of damage fillings
In power systems, reliable substation operation is critical for ensuring power supply stability and continuity. The reliability of substations directly affects the performance of the entire power system. However, existing substation monitoring systems have significant deficiencies in signal feature extraction accuracy and partial discharge (PD) type detection, which substantially limits the reliability and timeliness of fault diagnosis. To address these issues, this study proposes a hybrid approach based on ensemble empirical mode decomposition singular value entropy (EEMD-SVE). Moreover, a real-time PD monitoring system is developed using Hadoop and the hybrid algorithm. The suggested hybrid algorithm's performance is contrasted with that of alternative algorithms. Experimental outcomes revealed that the algorithm's F1 score was 97.88%, the average MSE value was 1.537, the average RMSE value was 0.462, and the fitting coefficient value was 0.968, all of which were better than the comparison algorithm. Subsequently, the application effect of the proposed monitoring system was analyzed. According to the findings, the system outperformed the comparison system in identifying four different types of partial discharges, with accuracy rates of 98.6%, 97.3%, 96.9%, and 97.8%, respectively. Furthermore, the study's suggested monitoring system outperformed the comparative system with an error rate of 0.14% and a throughput of 1132 requests per second. In conclusion, the algorithm and monitoring system proposed in this study are effective in enhancing the reliability of substation operation and fault diagnosis capabilities, providing a theoretical basis for partial discharge research.
Rolling bearing fault diagnosis faces severe challenges from data imbalance and variable operating conditions, restricting model generalization. We propose an Improved Generative Adversarial Network (IGAN) for high-fidelity fault sample synthesis. Core innovations are: 1) A 5-dimensional Composite Label Vector (CLV) that encodes physical information (load, fault diameter, location); 2) A robust conditional injection mechanism mapping labels to a high-dimensional space for precise guidance; 3) An Asymmetric Learning Rate (ALR) strategy for training stability. Comparative experiments on the CWRU dataset demonstrate that the proposed IGAN outperforms state-of-the-art baselines, boosting classifier accuracy to 99.1%. More importantly, the model synthesizes high-fidelity data for entirely unseen operating conditions via label vector interpolation and demonstrates strong generalization on the IMS natural run-to-failure dataset. This provides a generalizable, data-driven solution for few-shot, variable-condition fault diagnosis in realistic industrial scenarios.
