It is crucial to monitor the real-time and accurate status of an electric drive system and understand its interaction with driving behavior in order to meet the higher requirements for vehicle safety and reliability in the era of autonomous driving. Traditional wired sensors have limitations in system integration and energy autonomy. This study proposes a triboelectric nanogenerator (TENG) based on the centrifugal-force-enhanced contact mechanism, which can be directly integrated into the transmission shaft of electric vehicles to construct an intelligent self-powered monitoring system. This design effectively overcomes the bottleneck of unstable signals and insufficient durability of traditional rotating TENGs at high speeds by coupling centrifugal force and spring pre-tension and outputs stable and high signal-to-noise ratio sensing signals. On this basis, this study not only achieved high-precision real-time perception of the driving system speed but also further explored the rich information embedded in the centrifugal-force-enhanced contact TENG signal and extended it to the intelligent recognition of driving behavior and road conditions. Based on signal processing and the convolutional neural network and bidirectional long short-term memory model, the system has achieved fault diagnosis of key transmission component bearings in the transmission system with an accuracy rate of up to 96.1%. At the same time, the system can effectively recognize driving behaviors such as sudden acceleration and deceleration (recognition accuracy of 84%), as well as typical road conditions such as flat, slippery, and speed bumps (recognition accuracy of 89.9%), providing key information for automatic driving algorithm calibration and driving safety improvement. The self-powered embedded sensing technology developed in this study provides a new technological path for the efficient energy management and predictive maintenance system of intelligent connected vehicles and is a key sensing node for building future autonomous transportation systems.

This study presents a dynamic perception-weighted distillation approach for building an intelligent monitoring agent for fatigue testing machines, enabling real-time condition assessment and autonomous decision-making. The system integrates data acquisition, model training, service deployment, and decision modules, forming an end-to-end framework from data to agent level. A dynamic weighting mechanism guides the student model to emphasize high-value samples, improving fault recognition for complex scenarios. Through knowledge distillation and fine-tuning, the model achieves a balance between compactness and accuracy. After deployment, the agent autonomously collects operational data, diagnoses faults, and provides maintenance suggestions, continuously refining its judgments through interactive feedback. The proposed method enhances automation and intelligence, reduces dependence on expert experience, and improves the efficiency and reliability of equipment management.

Industrial defect detection is an essential component of modern manufacturing and plays a critical role in ensuring product quality, operational reliability, and production safety [...]

Harmonic reducers exhibit non-stationary and phase-dependent degradation behavior during long-term service, challenging the ability of classical stochastic degradation models to accurately assess reliability. To address phase-dependent differences in degradation behavior, this paper proposes a reliability assessment model based on a two-phase hybrid stochastic degradation process. In the proposed framework, the Wiener process is employed to characterize early-phase gradual degradation dominated by stochastic fluctuations, while the Inverse Gaussian process is used to describe later-phase monotonically accelerated degradation driven by cumulative damage. The framework allows for sample-level variability in transition times to more realistically capture individual degradation behavior. The Schwarz Information Criterion is also adopted to detect change points. Maximum likelihood estimation is performed for model parameter inference, and analytical expressions for the reliability function, cumulative distribution function, and probability density function are derived. Numerical results indicate that a change point exists for each tested product and that the proposed model achieves the best goodness of fit among the considered candidates, demonstrating its superiority in capturing phase-dependent characteristics of harmonic reducer degradation. In terms of reliability assessment bias, the proposed model (0.06%) significantly outperforms the Wiener degradation model (32%) and the IG degradation model (9.9%). These results further confirm that, under an identical failure threshold, the proposed approach yields more accurate and realistic reliability assessment outcomes.

As the manufacturing industry continues to advance its digital transformation, intelligent sensing technology has become a key support for achieving precise, efficient and automated quality control. However, current production line monitoring systems predominantly rely on fixed and costly monitoring equipment and sensors, lacking flexible and interactive first-person perspective perception approaches centered on on-site operators. Meanwhile, factory process monitoring often depends solely on visual expression rather than balancing the capabilities of the simulation model and visual state detection, leading to delayed responses to abnormal systems and hindering the adjustment strategy feedback. To address these limitations, this study provides wearable sensing for key workers, enriching the state perception capabilities in industrial scenarios. Furthermore, to achieve dynamic model and real-time visual representation of production line operations, a multi-source information-enhanced Petri nets model is proposed in terms of engineering and user-friendliness. With the solid mathematical basics of the Petri nets and the enriched human-machine data from the product line, this method provides an intuitive, dynamic and accurate reflection of the production system's real-time operational status, offering a scientific and reliable basis for operational decision-making. The proposed approach has been implemented in a real-world production system for reinforced concrete civil defense doors, and this engineering application can also be extended to many other scenarios.

Dynamic load spectra for electric-drive assemblies are difficult to estimate from road tests because the signals are non-stationary, non-Gaussian, and noisy. We propose a pseudo-damage-constrained data-model fusion framework that reconstructs torque/load histories while preserving rainflow counting and fatigue consistency. The approach combines trend extraction with nonlinear state estimation and an innovation-based adaptive step that enforces pseudo-damage equivalence to the raw signal within a controlled tolerance. Extreme-value fits are used only as tail diagnostics to verify that rare high-load behavior is preserved; they are not involved in cycle counting. On representative road data, the method achieved a Peak-Valley Preservation Rate approximate to 93% and the lowest weighted-MAPE (26.2%) among EKF, PF, KalmanNet, and LSTM baselines, with clear gains in fatigue-critical mid-high levels and no inflation of the spectrum tail. The results indicate that the proposed framework yields higher-fidelity spectra for durability analysis and test-bench replay while keeping established fatigue rules (four-point rainflow with Goodman correction) unchanged.

To address the challenge of maintenance decision-making for critical components in electro-hydraulic servo material fatigue testing machine, characterized by weak state observability and difficulty in degradation prediction, a multi-component joint maintenance decision-making method based on multi-head deep reinforcement learning is proposed. Considering the heterogeneity of the degradation mechanisms and observation methods for the four components—bearing beam, fixture, main machine sensors, and hydraulic oil tank—a continuous-discrete hybrid state Markov decision process (HS-MDP) is constructed. To account for differences in maintenance strategies across components, a differentiated discrete action space for each component is designed, and engineering feasibility constraints are explicitly integrated into the policy through action masking. A data-quality loss term, determined by the degradation level of the sensors, is introduced into the reward function to align the optimization objective with the metrological properties of the fatigue testing machine. Based on the Branching Dueling DQN framework, a Q-network structure is constructed, incorporating a shared encoder, an inter-component attention mechanism, and multi-head branched outputs. Taking a 100 kN electro-hydraulic servo fatigue testing machine as a case study, comparisons with baseline strategies such as periodic maintenance, threshold-based condition-based maintenance (CBM), independent DQN, and PPO indicate that the proposed method reduces the average annual total cost by 60.3% compared to periodic maintenance and by 42.6% compared to threshold-based CBM. The number of failures decreases from 9.8 times/year to 1.4 times/year, while data efficiency increases from 82.1% to 96.2%. Ablation experiments and robustness tests further verify the critical contributions of three key design elements: action masking, inter-component attention, and data-quality loss.

To address the insufficient prediction accuracy of multi-state parameters in electro-hydraulic servo material fatigue testing machines under complex loading and nonlinear coupling conditions, this paper proposes a multivariate sequence-to-sequence prediction model integrating a Long Short-Term Memory (LSTM) encoder, a Gated Recurrent Unit (GRU) decoder, and a multi-head attention mechanism. This approach enhances prediction accuracy and robustness across different control modes and load spectra by leveraging multi-channel inputs and cross-variable feature interactions, thereby capturing both short-term high-frequency dynamics and long-term slow drift characteristics. Experiments using long-term data from real test benches demonstrate that the model achieves a stable MSE below 0.01 on the validation set, with MAE and RMSE of approximately 0.018 and 0.052, respectively, and a coefficient of determination reaching 0.98. This significantly outperforms traditional identification methods and single RNN models. Sensitivity analysis indicates that a prediction stride of 10 achieves an optimal balance between accuracy and computational overhead. Ablation experiments validated the contribution of multi-head attention and decoder architecture to enhancing cross-variable coupling modeling capabilities. This model can be applied to residual-driven early warning in health monitoring, and risk assessment with scheme optimization in test design. It enables near-real-time deployment feasibility, providing a practical data-driven technical pathway for reliability assurance in advanced equipment.

The synthesis and preparation of supercapacitor cathode materials is one of the important ways to achieve the increase of the energy density of supercapacitors. CoMoO4 is a commonly used cathode material, but its own conductivity is poor, however, carbon can make up for this shortcoming of poor conductivity. In this paper, soluble starch with high solubility and uniform dispersion was used as the carbon source, and CoMoO4/carbon flower-like nanocomposites (CCMO) containing surface flocculants were successfully synthesized by a two-step hydrothermal reaction and applied as the cathode for supercapacitors, which showed excellent electrochemical properties. In order to determine the electrochemical properties of the prepared CCMO, it was tested in a three-electrode system. The test results demonstrate that the CCMO-3 electrode prepared under the condition of a quality ratio of CoMoO4 to soluble starch of 1:1 demonstrated the best electrochemical performance, with a specific capacitance of 415F g(-1). After 15000 cycles at a current density of 10 A g(-1), its specific capacitance remained at 87.5% of the initial reversible capacity. The supercapacitor device (CCMO//AC ASC) prepared with CCMO-3 and activated carbon, as both positive and negative electrodes, has an energy density of 12.44 Wh kg(-1) under a power density of 799.71 W kg(-1). At a current density of 5 A g(-1), it maintains 75% of its initial capacity after 10000 cycles. The excellent electrochemical performance of CCMO-3 shows its application potential in energy storage devices.

Accurately predicting the operating temperature of rotary kilns is crucial for their efficient operation and quality control. However, the lack of sufficient data for many rotary kilns hinders the accuracy of temperature prediction. The historical temperature data from rotary kilns operating under similar conditions often contain valuable information about temperature changes. To fully utilize this valuable information, this paper proposes the establishment of the LSTM-AE model, which combines the Long Short-Term Memory network and the Auto-Encoder. Additionally, a temperature prediction method for rotary kilns under short time samples is introduced by incorporating Transfer Learning and LSTM-AE. This method effectively transfers the knowledge of temperature variations from rotary kilns with similar working conditions to the target rotary kiln. By analyzing the performance of the method and comparing it with other data-driven methods, this study demonstrates that the method effectively mitigates the impact of rotary kiln temperature delay and abrupt variability. As a result, it significantly improves the accuracy of temperature prediction in rotary kilns with short time samples. The findings of this research provide a solid foundation for the development of temperature control strategies and the quality control in rotary kilns.

To accurately determine the early-failure time inflection point of inertial friction welders, this study collected and preprocessed 257 sets of fault data from 8 welders, and constructed a standardized dataset. Through Weibull Probability Plot (WPP) analysis, an inflection was observed in the data points. A two-segment Weibull model was thus adopted to replace the traditional single Weibull model, which effectively addresses the limitation that the single Weibull model cannot characterize the segmented failure characteristics (early-failure and randomfailure phases) of welders. Combined with maximum likelihood estimation using the Newton-Raphson algorithm and iterative calculation for residual sum of squares (RSS) minimization, the early-failure inflection point was determined to be 255.37 hours. Verification results show that before this time point, the welders are in the early-failure period with a decreasing failure rate; after this time point, they enter the random-failure period with a stable failure rate. This study provides a scientific basis for the operation and maintenance optimization of welders and early fault detection, and also offers a reusable quantitative analysis paradigm for determining the early-failure inflection point of similar equipment.

Photovoltaic systems are often exposed to complex environmental conditions, which can significantly affect their performance and lifespan. As the core component in converting solar energy to electrical power, photovoltaic (PV) panels may experience performance degradation due to operational wear and damage. Therefore, real-time monitoring systems are essential for ensuring the efficient operation of solar power systems and preventing potential energy loss. Efficient and reliable online monitoring of photovoltaic panel failures under complex conditions has always been a huge challenge. This study utilizes infrared thermography UAV to capture both surface and internal defect images of solar panels, constructing a dataset containing three common defect types. To mitigate background interference in large-scale images, we propose a Dynamically Adaptive and HighEfficiency Small Object Detection Network in Infrared Thermographic Images, enabling rapid and accurate extraction of multi-scale defects, even the small ones. A super-resolution weight matrix module is added to the network's frontend to enhance infrared thermography image resolution, filtering spatial and channel dimensions to extract defect regions and reduce computational complexity. A parallel efficient cross-channel fusion module refines multi-scale defect features, preventing information loss and enhancing feature extraction across scales. For small pitting defects, a dynamic local morphological dilation module is introduced to improve detection accuracy and reduce false positives. Lastly, a dynamic region proposal network adjusts the classifier and regressor based on prior detection results, improving overall detection accuracy. The model achieves a mAP of 98.1 %, outperforming others in speed and complexity, making it ideal for online defect detection applications.

Carbon/carbon composites serve as critical thermal protection materials in aerospace applications due to their exceptional high-temperature mechanical properties. Conventional performance evaluations inadequately replicate the coupled thermo-mechanical loading inherent to operational environments. Consequently, this study introduces a test system designed to simultaneously apply thermal ablation and biaxial tensile loading. Building on this, the study conducted biaxial tensile tests at 1500 degrees C ablation, revealing that the specimens' tensile strength is 59.3 MPa. The damaged specimens are analyzed using in situ observation devices, which calculated the average linear and volumetric ablation rates of the C/C composites to be 0.0274 mm s-1 and 0.0288 g s-1, respectively. Scanning electron microscopy shows grooves and oxides on the specimen surface, along with distinct features like matrix disappearance, exposed fiber tips, and sheath structures. Mechanistic analysis demonstrates that the synergistic effect of thermal-chemical ablation and biaxial stress accelerated interfacial degradation between fibers and matrix, leading to localized stress concentration and brittle fracture. The integrated methodology enables accurate performance assessment of thermal protection systems under realistic coupled-field conditions.

The reliability of turbine blades is vital under extreme operational conditions such as high temperature, high speed, and complex loads. To address the limitations of manual inspection and improve consistency and accuracy, we developed an automated fluorescent magnetic particle inspection (FMPI) system for real-time defect detection. This study proposes an Offset-enabled Multi-scale Pixel-wise Defect Detection (OMPD) model based on the TOOD framework. The model replaces the conventional CNN backbone with a deformable convolutional network (DCN) to better capture irregular defect shapes. An Adaptive Multi-scale Receptive Field (AMRF) module is introduced to dynamically adjust receptive fields for enhanced spatial feature extraction. Additionally, a Dilated Feature Aggregation and Pixel-level Perception Block (DFAPB) supplements shallow features with deep semantic information, improving edge localization. A lightweight boundary information supervision module is embedded into the FPN to preserve contextual details during feature fusion. Compared to the original TOOD model, OMPD improves AP from 71.2% to 74.9% and AP50 from 95.7% to 97.8%, while maintaining a high inference speed of 18.9 images per second. This meets the requirements for real-time industrial inspection with minimal computational overhead. The model achieves superior accuracy and robustness, particularly in detecting small and irregular defects, demonstrating its potential for practical deployment in automated turbine blade inspection systems.

The High-Temperature Biaxial Testing Apparatus (HTBTA) is a critical tool for studying the damage and failure mechanisms of heat-resistant composite materials under extreme conditions. However, existing methods for managing and monitoring such apparatus face challenges, including limited real-time modeling capabilities, inadequate integration of multi-source data, and inefficiencies in human-machine interaction. To address these gaps, this study proposes a novel digital twin-driven framework for HTBTA, encompassing the design, validation, operation, and maintenance phases. By integrating advanced modeling techniques, such as finite element analysis and Long Short- Term Memory (LSTM) networks, the digital twin enables high-fidelity simulation, real-time predictive modeling, and robust remote monitoring of HTBTA. The research contributes to bridging the knowledge gap in applying digital twin technology to high-temperature multi-axial testing systems. Unlike existing solutions, the proposed approach achieves <2% synchronization error, real-time monitoring with <100 ms delay, and predictive accuracy for temperature distributions under extreme conditions up to 2500 degrees C. The findings highlight the effectiveness of the digital twin in improving system reliability, enhancing interaction efficiency, and reducing maintenance costs. This study not only advances the application of digital twin technology in high-temperature material testing but also establishes a foundation for broader adoption in aerospace, automotive, and other industrial sectors. Future research directions include exploring non-proportional loading scenarios, expanding multi-environment simulations, and integrating in-situ observation techniques.

Electric drive assemblies operating under complex road conditions are often subjected to intense noise, nonstationary trends, and non-Gaussian disturbances, which pose significant challenges to accurate load spectrum construction. Particle filtering (PF) has been widely applied to address nonlinear and non-Gaussian problems, yet its performance is highly sensitive to the specification of process noise covariance, making fixed settings inadequate under time-varying conditions. Innovation-based adaptive estimation (IAE) improves robustness through covariance updating, but its heavy matrix operations impose prohibitive computational costs for embedded and realtime applications. To overcome these limitations, this paper develops a lightweight adaptive-covariance particle filtering method (LACU-PF). The method introduces a proportional update mechanism driven by normalized innovations to dynamically regulate process noise covariance, thereby retaining adaptability while substantially reducing computational complexity. Experimental validation using real-vehicle inertial measurement unit (IMU) data demonstrates that LACU-PF achieves superior spectrum fidelity and fatigue-related indices compared with conventional PF, while providing significantly higher efficiency than IAE. These results indicate that LACU-PF offers a feasible and scalable solution for real-time reliability assessment and embedded implementation of electric drive systems.

Defects on automotive paint surfaces not only compromise aesthetic quality but also diminish their protective capabilities. To address this issue, we propose the lightweight and multiscale feature enhancement You Only Look Once version 10 (LAM-YOLOv10) model, an advancement of YOLOv10s, specifically designed for rapid, accurate, and efficient detection of paint surface defects in industrial production line environments. The model integrates several innovative components: the dual-attention group normalization module, the triplet attention module, and the omni-dimensional dynamic convolution module. These elements facilitate defect detection through multiscale feature fusion, refined deep feature representation, and a lightweight overall network architecture. The LAM-YOLOv10 model achieves a mean average precision (mAP) at 50% intersection over union (mAP50) of 95.4% on the rearview mirror paint defect dataset, surpassing YOLOv10s by 0.9%. Validation on the NEU-DET dataset yields a mAP50 of 75.2%, indicating a 0.6% improvement over its predecessor. Notably, this model significantly reduces computational costs; floating-point operations and parameter counts are decreased by 20.8% and 11.3%, respectively. These optimizations render LAM-YOLOv10 highly suitable for online rapid defect detection within production lines where real-time performance and efficiency are paramount. The results demonstrate that LAM-YOLOv10 not only enhances detection accuracy but also provides a lightweight and efficient solution for industrial applications-enabling seamless integration into automated inspection systems with limited computational resources. (c) 2025 SPIE and IS&T [DOI:10.1117/1.JEI.34.3.033022]

The characteristics of multi-physics coupling in piezoelectric materials are influenced by the hygrothermal environment. Here, the thermo-hygro-mechanical-electro coupling isogeometric analysis method (THMEC-IGA) is proposed to explore the static and dynamic responses of piezoelectric structures. The inherent high-order continuity of non-uniform rational B-spline (NURBS) basis functions has higher computational accuracy than the traditional finite element method. Based on the B-spline theory, the classical piezoelectric theory and small strain theory, the constitutive equation of piezoelectric material THMEC-IGA and the displacement and electro potential shape functions are derived through the Gibbs free energy equation. The transformation method is used to accurately solve the THMEC problem, and the essential boundary conditions are appropriately applied at the control points. The subspace iteration method and Newmark method are used to solve the free vibration and transient response of the piezoelectric structures in the hygrothermal environment. The superiority of THMEC-IGA in the THMEC problem is verified using numerical examples. The advantages of THMEC-IGA in solving multi-physical field coupling problems are verified in comparison with the finite element results. The influence of empirical constants and hygrothermal changes on the mechanical properties of piezoelectric structures is studied, providing a reference for the design and analysis of piezoelectric structures.