Accurate respiratory rate (RR) measurement is essential for early detection of physiological deterioration, yet conventional approaches remain limited by poor robustness and restricted suitability for continuous or remote monitoring. ECG-derived respiratory rate (EDRR) offers a non-invasive alternative by exploiting respiration-induced modulations in cardiac electrical activity, including morphological changes, heart rate variability, and their combined effects. This review provides a structured synthesis of EDRR methods, spanning physiological mechanisms, signal acquisition and preprocessing considerations, and algorithmic approaches including morphology-based, autonomic (heart rate variability/spectral), and fusion strategies. We further evaluate datasets, validation practices, and performance trade-offs across controlled and real-world conditions. Key challenges include susceptibility to motion artifacts, inter-subject variability, inconsistent evaluation protocols, and limited generalizability in ambulatory settings. Addressing these limitations is critical for translating EDRR techniques into robust, scalable solutions for wearable and telehealth-based respiratory monitoring.

Local field potential (LFP) decoding is critical for the clinical translation of intracortical brain-machine interfaces (iBMIs), yet existing decoding methods are limited by three key bottlenecks: insufficient single-scale feature utilization, inefficient multi-scale feature fusion, and poor robustness across task paradigms and chronic recording conditions. To address these challenges, we propose Dual-VCT, a novel dual-branch Variational Mode Decomposition-Convolutional Neural Network-Transformer (VMD-CNN-Transformer) model for end-to-end LFP decoding. The core innovation of Dual-VCT is its symmetric time-frequency parallel architecture with independent VMD modules embedded in both branches: a temporal branch decomposes local motor potential (LMP) signals via VMD to capture motion-related instantaneous neural activity, while a frequency-domain branch leverages VMD to isolate task-relevant spectral power components, with a hierarchical fusion pipeline enabling robust cross-scale feature integration. Validated in non-human primate experiments, Dual-VCT achieved a classification accuracy of 0.930±0.023 in the 3-class spatial grasping task, and a Pearson correlation coefficient (CC) of 0.910±0.023 in the finger point-to-point tracking task. It significantly outperformed all comparative dual-branch methods under identical experimental conditions (p < 0.05), delivered a 4% performance gain over single-feature decoding, and exhibited strong cross-task robustness and cross-day stability. Ablation experiments confirmed the core contribution of the dual-branch VMD design. This work provides a high-performance structured paradigm for LFP decoding, with a clinically oriented design that supports the long-term stability of chronic iBMI systems.

Two-dimensional MXene materials are promising for flexible electronics due to their high conductivity and mechanical strength, yet their films often suffer from poor mechanical robustness and a trade-off between conductivity and mechanical performance. This study introduces a bioinspired "sequential bridging" strategy to fabricate PDA/MXene/SA/Ca2+ films (PMSC) via synergistic polydopamine (PDA) surface modification, sodium alginate (SA) hydrogen-bond crosslinking, and Ca2+ ionic bridging. The film exhibits exceptional comprehensive properties, achieving a tensile strength of 256.64MPa, a Young's modulus of 8.06GPa, and a toughness of 8.39MJ/m3, which are 7.8, 3.2, and 10.9 times higher than those of pure MXene film, respectively, while retaining a high electrical conductivity of 1700.91 S/cm. Systematic characterization analysis identified multiple enhancement mechanisms, including PDA surface modification, SA induced hydrogen-bond crosslinking, and Ca2+ mediated ionic-bond energy dissipation. This multi-crosslinking approach overcomes the limitations of single network strategies, offering a scalable route to high-performance flexible electronics.

This study addresses the inherent limitations of conventional PI control in speed regulation systems, particularly its poor robustness. A dual‐loop model predictive control (MPC) strategy integrating speed and current loops is proposed, treating the motor as a multi‐input multi‐output system to effectively mitigate issues such as weak robustness and slow dynamic response during operation. The speed loop employs Model Predictive Speed Control (MPSC) for speed regulation. However, since MPC performance partially relies on the accuracy of the motor's mathematical model, load disturbances may degrade control performance. To address this, a load observer is introduced for the LIM speed loop to estimate and compensate for disturbances, significantly improving control accuracy. Meanwhile, the current loop employs model predictive flux control (MPFC), enhanced by an improved generalized dual‐vector control strategy to optimize steady‐state performance, thereby effectively reducing current ripple and torque pulsation. Simulation results demonstrate that the proposed dual‐loop MPC strategy exhibits superior performance under speed step changes and load disturbances, featuring minimal overshoot, fast dynamic response, high steady‐state accuracy, and strong anti‐interference capability. Additionally, the observer effectively tracks the given load. This study provides a practical control solution for high‐performance LIM applications in rail transit and industrial automation. © 2026 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

Breath-based sensing has become a promising non-invasive method for early disease screening, especially for chronic metabolic disorders such as diabetes mellitus. Although elevated breath acetone is a known marker of impaired glucose metabolism, current breath-based sensing methods are mostly based on single-biomarker sensing or multiple sensor arrays, so they suffer from poor robustness and complexity. In this paper, an artificial intelligence (AI) assisted functionalized microcantilever sensor with preliminary experimental validation for non-invasive breath-based screening for diabetes using volatile organic compound (VOC) fingerprint detection is proposed. The proposed sensor is based on a single microcantilever coated with multiple VOC-responsive functional layers and can selectively interact with host-related VOCs associated with diabetes. Adsorption-induced mass loading and surface stress variations are responsible for unique mechanical response patterns that act as disease-specific fingerprints. Analysis shows resonant frequency shifts of the order of tens of hertz and static deflections in the nanometer scale for different breath compositions of VOCs. Furthermore, AI-aided mechanical fingerprint classification achieved up to 96.8% accuracy under controlled evaluation conditions.

To overcome offline lag, difficult decoupling, poor real-time accuracy &amp; poor adaptability of geometric errors in large CNC long-stroke machining, this paper proposes a multi-source fusion intelligent compensation system. For an XYTZ gantry CNC, it builds a multi-source network (dual-frequency laser interferometer, photoelectric encoder, industrial gyroscope, temperature sensor); based on HTM &amp; multi-body theory, establishes a 21-term geometric error model integrating Abbe &amp; Bryan principles &amp; friction thermal deformation for geometric-thermal coupling; uses adaptive Kalman filtering for real-time multi-source data fusion &amp; error sensitivity matrix for fast spatial error decoupling; and deploys a ROS2-based distributed real-time compensation architecture delivering ms-level compensation via FANUC's EPO interface synchronized with the interpolation cycle. Theory &amp; design show: over 5m stroke, compensation accuracy &gt;±10 μm, delay ≤1.5 ms, solving weak real-time &amp; poor robustness, thus supporting precision large-CNC machining.

Hyperspectral unmixing aims to estimate endmember spectra and their corresponding abundance fractions at the subpixel scale, which is a critical preprocessing step for quantitative analysis of hyperspectral remote sensing imagery. While deep learning-based methods have achieved remarkable progress, three fundamental challenges remain: (i) reliance on a single shared spatial prior that cannot decouple the heterogeneous spatial patterns of different land covers; (ii) the lack of synergy in jointly optimizing endmember extraction and abundance estimation; (iii) the poor robustness of unsupervised training to complex mixtures, noise, and class imbalance. To address these issues, we propose a novel unsupervised unmixing framework that integrates adaptive orthogonal multi-faceted graph representation with curriculum learning. Specifically, we design an Adaptive Orthogonal Multi-Faceted Graph Generator (AOMFG) to learn a set of independent orthogonal graph structures, achieving spatially informed decoupling of land cover patterns. Then, a dual-branch collaborative optimization network is constructed: a Graph Convolutional Network (GCN) branch that incorporates the learned spatial topological priors for abundance estimation, and a 1D Convolutional Neural Network (1DCNN) branch that employs a query-attention mechanism to adaptively aggregate pure spectral features for endmember extraction. Finally, we introduce a three-stage curriculum learning strategy that progressively fine-tunes the model, which significantly enhances its performance. Extensive experiments on three widely used real-world benchmark datasets demonstrate that our proposed framework consistently outperforms state-of-the-art methods in both endmember extraction and abundance estimation accuracy. Comprehensive ablation studies, parameter sensitivity analysis, and noise robustness tests further validate the effectiveness of each core component.

While observational studies have identified a link between dietary factors and the incidence of stroke, whether this association reflects a causal relationship remains poorly defined. For this work, we utilized a two-sample Mendelian randomization (MR) method to investigate the potential causal association between dietary intake and the risk of stroke. Genetic instrumental variables were extracted from the IEU OpenGWAS and GWAS Catalog databases. The inverse variance weighted method was used to calculate the MR estimates, and sensitivity analyses were performed to evaluate the robustness of the results. Dried fruit intake was associated with a reduced risk of stroke (odds ratio [OR] = 0.991, 95% confidence interval [95% CI]: 0.983-0.998, P = .013) and its subtypes, including ischemic stroke (OR = 0.634, 95% CI: 0.417-0.965, P = .034), small-vessel ischemic stroke (OR = 0.381, 95% CI: 0.195-0.743, P = .005), and lacunar stroke (OR = 0.320, 95% CI: 0.149-0.686, P = .003). Oily fish intake exerted a protective effect against stroke (OR = 0.994, 95% CI: 0.988-0.999; P = .016). Non-oily fish intake reduced the risk of lacunar stroke (OR = 0.256, 95% CI: 0.081-0.807, P = .020), and pork intake reduced the risk of intracerebral hemorrhage (OR = 0.169, 95% CI: 0.031-0.930, P = .041). No significant causal associations were found between stroke and the intakes of processed meat, salad/raw vegetables, fresh fruits, beef, cooked vegetables, poultry, lamb/mutton, or coffee. Reverse MR analyses revealed no widespread and robust reverse causal associations; only 2 nominally significant signals were detected, which had poor analytical robustness or negligible clinical relevance and did not affect the core forward causal inferences. This study provides genetic evidence supporting the causal relationships between specific dietary factors (dried fruit, oily fish, nonoily fish, and pork) and stroke risk, which may have implications for dietary guidance and stroke prevention strategies.

Traditional pedestrian inertial navigation (PDR) algorithms usually assume that the carrying mode of a smartphone is fixed and remains horizontal, while ignoring the significant impact of dynamic changes in the carrying mode on heading estimation, which is the core element of PDR algorithms. In practical application scenarios, pedestrians often change their way of carrying smart terminals (e.g., calling) according to their needs, corresponding to the difference in the heading estimation method; especially when the mode is switched, it will cause a sudden change in heading, which will lead to a significant increase in the localization error if it cannot be corrected in time. Existing smart terminal carrying mode recognition methods that rely on traditional machine learning or set thresholds have poor robustness; lack of universality, especially weak diagnostic ability for mutation; and can not effectively reduce the heading error. Based on these practical problems, this paper innovatively proposes a PDR framework that tries to overcome these limitations. Based on this research purpose, firstly, this paper classifies four types of common carrying modes based on practical applications and designs a CNN-LSTM hybrid model, which can classify the four common carrying modes in near real-time, with a recognition accuracy as high as 99.68%. Secondly, based on the mode recognition results, a multi-layer heading correction strategy is introduced: (1) introducing a quaternion-based universal filter (VQF) algorithm to realize the accurate estimation of initial heading; (2) designing an algorithm to accurately detect the mode switching point and developing an adaptive offset correction algorithm to realize the dynamic compensation of heading in the process of mode switching to reduce the impact of sudden changes; and (3) considering the motion characteristics of pedestrians walking in a straight line segment where lateral displacement tends to be close to zero. This study designs a heading optimization method with lateral displacement constraints to further inhibit the drifting of the heading caused by the slight swaying of the smart terminal. In this study, two validation experiments are carried out in two different environment-an indoor corridor and a tree shelter-and the results show that based on the proposed multi-layer heading optimization strategy, the average heading error of the system is lower than 1.5°, the cumulative positioning error is lower than 1% of the walking distance, and the root mean square error of the checkpoints is lower than 2 m, which significantly reduces the positioning error and shows the effectiveness of the framework in complex environments.

To address the issues of weak inter-frame motion correlation and poor recognition robustness in video-based pig behavior recognition, as existing methods fail to fully exploit the spatiotemporal dynamic features of skeletons and can hardly capture fine behavioral details, this study proposes a skeleton-based spatiotemporal dynamic modeling method for pig behavior recognition. We use DeepLabCut (DLC) to accurately extract pig skeleton keypoints and construct the topological structure, streamline the ST-GCN by removing redundant network layers, and design an improved BCST-GCN model with a global-local self-attention BC module to dynamically reconstruct topological correlations, so as to effectively capture non-physical connections and complex spatiotemporal behavior characteristics. Experimental results show that the proposed framework can effectively recognize typical behaviors such as feeding, walking, lying, and dog-sitting posture, and the improved model yields 6.94%, 5.61%, and 6.88% increments in accuracy, precision, and recall respectively compared with the baseline model. The proposed method achieves accurate and efficient pig behavior recognition, solves the problems of weak temporal correlation and insufficient feature extraction in traditional models, and provides a reliable technical solution for intelligent monitoring in pig farming scenarios, supporting the intelligent upgrading of the breeding industry.

ABSTRACT This paper investigates adaptive integral sliding mode control (AISMC) for stochastic nonlinear systems based on Takagi–Sugeno (T–S) fuzzy modeling. To address this issue, an adaptive integral sliding mode control scheme is proposed. Specifically, adaptive parameter estimation is incorporated into the fuzzy controller to mitigate the conservative control performance and poor robustness associated with traditional fixed‐gain designs. Based on stochastic Lyapunov stability theory, it is proven that the trajectories of the closed‐loop system inevitably remain on the integral sliding surface. The stability of the sliding motion is verified via linear matrix inequalities. Finally, simulations of an inverted pendulum example demonstrate the superiority and effectiveness of the proposed method.

ABSTRACT Industrial Internet of Things (IIoT) networks face the critical challenge of balancing data privacy protection with network efficiency in multidevice, multinode environments. Existing methods often address privacy or efficiency issues in isolation, leading to limitations such as high computational overhead, poor robustness, and insufficient explainability in large‐scale deployments. To address these challenges, this paper proposes a collaborative optimization method that integrates Explainable Artificial Intelligence (XAI) with data privacy protection. The core of the proposed method includes: designing an XAI‐guided privacy protection model to enhance transparency, constructing a collaborative optimization framework for privacy and scheduling, and employing reinforcement learning for dynamic resource scheduling. Experimental results show that the proposed method achieves a latency of 15.2 ms and a bandwidth utilization of 92.3%, outperforming baseline methods such as Deep Q‐Network (DQN) and Shortest Job First (SJF), while also reducing privacy risk penalty (0.48) and improving system robustness. This study offers a novel approach for efficient and secure management of IIoT networks and provides a reference for privacy‐aware network optimization.

Noble‐metal nanoframes are attractive for catalytic applications due to their open structures and large specific surface areas. However, traditional nanoframes are known to suffer from poor structural robustness because of their ultrathin ridges. Here, we address this issue by developing a new class of nanoframes characterized by relatively thick but hollow ridges. Our demonstration is based upon Pd and a typical synthesis involves the one‐shot injection of PdBr 2 powder suspension into an aqueous mixture containing Ag nanocubes, ascorbic acid, poly(vinyl pyrrolidone), and KBr held at 75°C. The poor solubility of PdBr 2 in water enables slow release of Pd (II) and thus a steady reduction rate for controlling the deposition of Pd atoms. During the synthesis, the side faces of Ag nanocubes are continuously carved away through galvanic replacement, accompanied by the deposition of Pd atoms on the corners and edges for the generation of cage cubes—cubes with through holes across the opposite faces. At a later stage, the Ag remaining in the ridges is etched away, leading to the formation of hollow ridges with a final composition of Pd 1 Ag 2 . The Pd–Ag alloy nanoframes exhibited good structural robustness during 1 h of chronoamperometric test toward formic acid oxidation reaction.

Radiomics has shown potential for quantitative characterization of tumors in molecular imaging; however, its clinical translation in theranostic 1 7 7Lu SPECT/CT remains limited due to poor robustness of extracted features to reconstruction variability and partial volume effects. Establishing reproducible radiomics biomarkers across correction strategies is therefore a prerequisite for reliable clinical modeling and treatment monitoring. This study aimed to evaluate radiomics feature reproducibility, defined as the stability of feature values across different partial volume correction (PVC) strategies and reconstruction settings, in clinical 1 7 7Lu SPECT/CT imaging. In addition, we explored two volumetric shape-based indices, the metastasis-to-liver ratio (MLR) and metastasis-to-spare liver ratio (MSLR), as surrogate markers of hepatic metastatic burden in the theranostic treatment setting. In 13 patients (40 scans) treated with 177Lu, 837 radiomics features were extracted from 11 abdominal regions and metastases on SPECT/CT using original and wavelet-decomposed images across four bin widths (50-200). Two post-reconstruction PVC methods, namely Richardson-Lucy (RL) and Reblurred Van Cittert (RVC), were applied. Feature reproducibility was quantified using two complementary metrics: the intraclass correlation coefficient (ICC) to assess feature-level stability across PVC strategies, and the concordance correlation coefficient (CCC) to evaluate pairwise agreement and systematic bias among reconstruction methods. Visual image quality assessments were independently performed by two experienced nuclear medicine specialists in a blinded setting. Exploratory metastatic tumor burden was assessed descriptively using 3D shape-based MLR and MSLR indices. Low-frequency wavelet decomposition (LLL-wavelet) and original features showed the highest reproducibility (ICC≥0.90 in>95% of liver and metastasis features at BW50), whereas high-frequency features and larger bin widths demonstrated reduced stability. CCC analysis revealed excellent agreement between RL and RVC (≥0.95 in major organs at BW50-100), while agreement with uncorrected SPECT (no PVC) was consistently lower, especially for high-frequency features. RL achieved higher visual scores in sharpness and contrast (p<0.01), with good inter-reader agreement supporting the consistency of these assessments. MLR/MSLR demonstrated inter-patient variability and were explored descriptively as indices of metastatic liver burden. Reproducibility in theranostic SPECT radiomics is highly feature- and organ-dependent and is further influenced by scanner-specific factors and reconstruction protocols, which remain critical for real-world clinical translation. RL and RVC showed stronger mutual agreements than each with uncorrected SPECT. Importantly, only RVC translated visual improvements into enhanced feature-level reproducibility, while RL provided the most consistent overall balance of reproducibility and image quality, supporting its role as the preferred PVC strategy for clinical and modeling applications. Robust radiomics feature selection as well as standardized reproducible PVC strategies are essential to generate methodological harmonization for future clinical translation and to support integration of radiomics analyses into personalized SPECT theranostics.

To address the low accuracy and poor robustness of anomaly detection for planetary gear trains under asymmetric sample conditions, this paper proposes a generalized generative adversarial network (GGAN) method. The proposed approach integrates a generative adversarial network, an autoencoder, and a contrastive learning mechanism. By leveraging a multi-scale discriminator and a residual network to extract nonlinear feature discrepancies, and combining kernel density estimation to quantify anomaly probabilities, it significantly improves anomaly detection accuracy. Building on the GGAN-based anomaly detection method, a three-stage anomaly evaluation framework is developed. In the normal operation stage, the initial detection model is trained using only normal samples. In the early degradation stage, the detection threshold and model parameters are refined using a limited number of anomalous samples. In the severe degradation stage, multi-scenario data fusion and similarity analysis are incorporated to achieve robust evaluation. Experimental results demonstrate that the proposed method can provide theoretical support and practical reference for research and applications in related fields.

Agricultural thermal stress and water loss are critical for sustainable food security in increasingly hot and arid climates, exacerbated by global warming. Passive radiative cooling, leveraging sunlight reflection and deep-space infrared emission, offers a promising solution for agricultural thermal management. However, conventional petrochemical-based radiative cooling materials face environmental challenges from persistent waste, while emerging biomass alternatives often suffer from inadequate cooling performance, poor mechanical robustness, and limited durability. Here, we engineered a robust, sustainable, and recyclable radiative cooling film via a vacuum-assisted hierarchical self-assembly process. This biocomposite features a heterogeneous interlocking structure, which synergistically achieves exceptional solar reflectance (98.1%) and infrared emittance (93.2%), enabling subambient cooling. Field tests demonstrate significant average temperature reductions of 4.6 and 2.4 °C relative to bare soil and commercial reflective mulch, concurrently suppressing soil evaporation and enhancing crop yield. Our material exhibits superior tensile strength (38.7 MPa), significantly outperforming standard commercial reflective mulch (14 MPa). Critically, we introduce a mild water-mediated recycling process achieving near-complete material recovery (>90%), eliminating end-of-life waste accumulation and its detrimental impact on agricultural soils. Combined with exceptional UV resistance, flexibility, and sustainability, this high-performance, recyclable radiative cooler presents a transformative approach for sustainable agricultural thermal management.

Agentic AI has emerged as a promising paradigm for autonomous reasoning and execution in complex AI-driven applications; however, its effective deployment in cloud-native environments remains challenging due to the lack of unified platform architectures that jointly support task decomposition, multi-agent collaboration, and adaptive cloud resource orchestration. In practical scenarios such as automated data analytics, AI DevOps, and MLOps pipelines, Agentic AI systems must operate over dynamic containerized infrastructures where resource availability, execution cost, and failure conditions continuously change. Existing approaches typically decouple agent-level decision making from cloud-native scheduling, resulting in limited scalability and poor robustness. To address these limitations, this paper proposes CANAO, a Cloud-Aware Native Agentic AI framework for adaptive task orchestration in cloud-native environments. CANAO models complex AI workloads as dynamically reconfigurable task dependency graphs and enables coordinated collaboration among Planner, Executor, and Critic agents. By incorporating real-time cloud resource awareness into the agent orchestration loop, CANAO supports adaptive scheduling, partial task re-planning, and self-healing execution on Kubernetes-based platforms. A prototype system is implemented using cloud-native technologies and evaluated on representative automated data analysis and AI DevOps workflows. Experimental results show that CANAO significantly outperforms baseline orchestration methods under dynamic cloud conditions. Compared with static DAG-based scheduling, CANAO reduces end-to-end task execution time by approximately 34.3% and cloud resource cost by nearly 30%, while lowering the task failure rate by over 34%. These improvements demonstrate the effectiveness of cloud-aware agent collaboration and adaptive task orchestration in large-scale cloud-native AI workflows.

Automated cytogenomic analysis has long been limited by narrow task scope, high annotation demands, and poor robustness to real-world complexity. Here, we introduce CHROMA, the first single-chromosome foundation model for cytogenomics that enables comprehensive, cell-level detection of a wide spectrum of chromosomal abnormalities-including both common and ultra-rare types-in a single, unified framework. Pre-trained on over 4 million chromosomal images from more than 84,000 specimens using self-supervised learning, CHROMA achieves robust and comprehensive detection of numerical and structural abnormalities across diverse classes, dramatically reducing expert annotation workload by 40% through efficient label utilization. The model maintains state-of-the-art accuracy even under highly imbalanced data and challenging imaging conditions, supporting reliable deployment as a risk-aware screening and triage tool, particularly in settings with limited expert availability. An integrated risk-control strategy further ensures safe application by automatically flagging uncertain or rare cases for expert review. By bridging foundational AI advances with real-world clinical needs, CHROMA paves the way for scalable, accessible, and precise cytogenomic analysis in both advanced and underserved healthcare environments.

Abstract Wide-bandgap perovskite solar cells (WBG-PSCs) are indispensable top subcells in perovskite-based tandem solar cells targeting power conversion efficiencies (PCEs) beyond 35%. Their performance, however, is limited by the strong thickness dependence of the absorber and the poor thermal robustness of conventional hole-transport layers (HTLs). Here, we theoretically explore the potential of 2,3-thienoimide-ended hexyl-substituted quaterthiophene (HT4N), an oligothiophene derivative, as an efficient HTL in WBG-PSCs. Compared to conventional HTLs, HT4N offers significant advantages due to its superior hole mobility and deep HOMO level, which enhance charge extraction and interface performance by improving energy-level alignment at the perovskite/HTL interface. Integrating a 40 nm HT4N interlayer with 2,2′,7,7′-tetrakis(N,N-di-p-methoxyphenylamino)-9,9′-spirobifluorene (Spiro-OMeTAD) in a dual-HTL configuration, combined with a 600 nm-thick perovskite absorber, achieves a champion PCE of 24.16% and an open-circuit voltage ( V OC ) of 1.37 V, surpassing the single-HTL device (21.69%). The HT4N layer effectively suppresses interfacial recombination, reduces parasitic absorption in Spiro-OMeTAD, and improves thermal stability, with the dual-HTL device retaining &gt;97% of its initial PCE at 400 K, while single-HTL devices degrade to ∼80%. Extending this approach to semitransparent WBG-PSCs using indium zirconium oxide as the transparent electrode achieves a PCE of 22.30%. These findings establish HT4N as a promising hole-transport material and demonstrate that dual-HTL engineering offers a powerful route to highly efficient, thermally robust, and tandem-compatible WBG-PSCs.

The meshing contact performance of spiral bevel gears is critical for transmission accuracy and service life but is inevitably influenced by manufacturing deviations. Existing tooth contact analysis (TCA) and lubrication-related studies for spiral bevel gears are mostly based on ideal theoretical tooth surfaces, failing to reflect the actual meshing state of as-machined gears with inherent machining deviations, and have poor robustness for complex deviated spatial surfaces. To accurately assess the actual meshing state, this paper proposes a novel contact performance analysis method based on a high-precision digital tooth surface reconstructed from one-dimensional probe measurement data. Unlike traditional TCA methods that rely on complex principal curvature calculations, this approach eliminates the mounting distance parameter by simplifying the meshing coordinate system, and employs a variable-radius cylindrical cutting method combined with a binary search algorithm to determine the instantaneous contact ellipse, effectively reducing computational complexity and improving solution robustness for deviated tooth surfaces. Experimental validation demonstrates that the digital tooth surface achieves a reconstruction accuracy of 2.6 × 10−5 mm. Furthermore, the method accurately predicts the contact pattern location and transmission error, with a discrepancy of only 4.7% compared to theoretical design values, which is highly consistent with the no-load rolling test results. This study confirms that the proposed method effectively reflects the actual meshing condition of machined gears, providing a practical theoretical foundation for the high-quality manufacturing and control of spiral bevel gears. Meanwhile, the high-fidelity contact characteristics of as-machined tooth surfaces output by this method can provide reliable input boundaries for thermoelastohydrodynamic lubrication (TEHL) simulation, friction loss prediction and anti-scuffing design of spiral bevel gears considering machining deviations.