Search Results (359)

The Internet of Medical Things (IoMT) is a promising technology for enabling smart and efficient healthcare systems through continuous physiological monitoring, early anomaly detection, and proactive patient management. However, IoMT sensors are vulnerable to False Data Injection Attacks (FDIAs), in which adversaries can manipulate sensor measurements to compromise diagnostic accuracy, mislead clinical decision-making, and threaten patient safety. Existing detection approaches often rely on population-level statistical models that may not fully capture individual physiological variations or residual-based thresholds designed for relatively simple attack scenarios, limiting their ability to exploit the spatio-temporal dependencies of multi-sensor physiological streams and detect stealthy or adversarial FDIAs. This paper proposes a patient-specific FDIA detection framework based on a Graph Convolutional Network–Gated Recurrent Unit (GCN–GRU) digital twin that learns an individual patient’s normal physiological behaviour from clean baseline telemetry. The trained digital twin is integrated into a Kalman filter as the state prediction model, and the resulting standardised innovation residuals are used as detection features. To characterise stealthy attack behaviours, four complementary window-based feature groups are extracted from the innovation sequence: innovation statistics, sensor correlation drift, temporal smoothness, and uncertainty mismatch. A CNN-1D classifier is then trained to learn discriminative temporal attack patterns from these features for accurate detection. A structured attack taxonomy comprising five stealthy and adversarial FDIA scenarios is developed, where attacks are injected as smooth gradual or abrupt coordinated modifications to sensor measurements while remaining within plausible physiological ranges. Experiments conducted on the WUSTL-EHMS-2020 benchmark dataset demonstrate that the proposed framework achieves an F1-score of 94.3%, outperforming Isolation Forest and PCA Reconstruction by 34 percentage points. Furthermore, the proposed framework reduces the false alarm rate to 3.6%, compared with 35.1% and 9.2% achieved by Isolation Forest and PCA Reconstruction, respectively. These results demonstrate the effectiveness of the proposed framework for reliable detection of stealthy FDIAs in IoMT-based healthcare systems. Full article

Background: Anorexia nervosa (AN) is associated with extensive biochemical alterations at the time of clinical presentation. Despite distinct behavioural profiles, it remains unclear whether the restricting (AN-R) and binge–purge (AN-BP) subtypes differ in their biochemical profile at residential admission, a question with direct implications for clinical monitoring protocols. Methods: We conducted a retrospective cross-sectional study of 177 consecutive AN inpatients (AN-R: n = 130; AN-BP: n = 47; 98.3% female; mean BMI 15.8 ± 2.0 kg/m²; mean age 19.2 ± 5.5 years; 53.7% adolescents) admitted to a specialist residential unit (2018–2026). Thirty-five biochemical parameters were assessed. Unadjusted comparisons used Mann–Whitney U tests with rank-biserial effect sizes; multivariable logistic regression adjusted for BMI, age, and illness duration. Results: Overall, 98.3% of patients presented at least one biochemical anomaly (mean 4.4 ± 2.6 per patient). The most frequently abnormal parameters were LDH elevation (75.0%), vitamin D insufficiency (73.0%), low T3 syndrome (defined as low free triiodothyronine [FT3]) (58.0%), FT4 suppression (53.5%), ALT elevation (35.1%), and leucopenia (32.3%). AN subtype predicted none of these outcomes in unadjusted or multivariable analyses (all p > 0.10). The odds of presenting ≥3 simultaneous anomalies were not significantly associated with subtype, BMI, or illness duration in multivariable analysis. In contrast, older age independently predicted greater odds of low T3 syndrome (OR = 1.14 per year, 95% CI [1.00, 1.30], p = 0.045) and leucopenia (OR = 1.22 [1.06, 1.39], p = 0.004). No differences were observed between adolescents and adults in unadjusted comparisons. Conclusions: Despite their distinct behavioural profiles, no statistically significant biochemical differences were detected between AN-R and AN-BP at residential admission. These findings support routine standardised biochemical screening in all patients with anorexia nervosa admitted to residential care, regardless of behavioural subtype. Older age may identify a subgroup warranting closer endocrine and haematological monitoring. Full article

Unmanned surface vehicle (USV) formations operating under GNSS denial require accurate relative localization using proprioceptive sensors and inter-vehicle ranging. This paper presents an anomaly-aware adaptive extended Kalman filter for four-USV formations using inertial measurements, compass, and ultra-wideband ranging, and systematically characterizes its operating envelope. Observability analysis establishes that S-curve maneuvering achieves structural rank 24, with only global translation unobservable, while straight-line motion leads to a rank deficiency of exactly seven dimensions All four gyroscope biases remain observable under both trajectories. The proposed filter integrates chi-square testing, cumulative sum (CUSUM) detection, and bias drift rate monitoring to trigger coordinated R adaptation and Q-boost mechanisms. Controlled experiments spanning outlier magnitudes and drift rates reveal three performance regimes, clean conditions with equivalent performance across all variants, moderate outliers $[3{\sigma }_d,10{\sigma }_d]$ where the proposed method achieves 4.8–13.4% improvement, and extreme outliers where all robust methods converge. Critically, pure bias drift experiments expose a structural limitation of single-hypothesis, residual domain robustification within the tested drift range—all variants exhibit equivalent performance across the tested drift rates, analytically attributable to Kalman gain partitioning that distributes innovations between position and bias subspaces. The characterized operating envelope establishes that robust mechanisms provide measurable benefits for transient anomalies but encounter hard boundaries under persistent drift conditions, with all variants converging to equivalent performance across the tested range, necessitating multi-hypothesis or constraint-based approaches. Full article

Traffic inside a cloud Virtual Private Cloud (VPC) is filtered by two layers with fundamentally different semantics: a stateless, ordered, first-match Network ACL (NACL) and a stateful, unordered, allow-only Security Group (SG). Existing analyzers decide point-to-point reachability using satisfiability solvers, Datalog engines, or binary decision diagrams, but do not identify, at the rule level, which rules are dead, redundant, or dominated, nor explain why. We provide a closed-form set algebra over the two layers. Representing each rule field by its boundaries makes a rule a hyper-rectangle and a layer a union of boxes; the effective admitted region Φ = A(N) ∩ A(G) is then a finite union of disjoint boxes computable from rule endpoints alone. We define a taxonomy of single- and cross-layer anomalies—shadowed NACL rules, dead SG rules, Φ-redundant rules, Φ-ineffective NACL allows, and layer disagreement—characterize each by a decidable region predicate, and prove an exact iff-condition for SG Φ-redundancy. A boundary-only detection algorithm is sound and complete for the exactly decidable anomaly classes, running in O((k + t)^d) time for fixed dimension d, and the disjoint box decomposition of Φ gives a minimal anomaly-free form that is unique up to merging adjacent coplanar boxes. A single-file implementation matches brute force on millions of packets, staying orders of magnitude below the worst-case bound; the parametric model extends unchanged to IPv6 and ICMP. Full article

Slab reheating furnaces are among the most energy-intensive units in hot-rolling production lines, and their operating states directly affect fuel consumption, temperature uniformity, and production cost. High energy consumption conditions often arise from coupled changes in reheating time, furnace-temperature regulation, combustion status, and production rhythm, making them difficult to identify using fixed energy thresholds or manual experience alone. This study proposes an energy-efficiency-constrained DDPM-ConvTransformer method for data-driven high-energy-consumption anomaly diagnosis in slab reheating furnaces. Continuous industrial records are converted into fixed-length production windows, and a denoising diffusion probabilistic model is used to learn the multivariate temporal distribution of normal operating conditions. A convolutional module and a Transformer encoder are embedded in the denoising network to capture local thermal-process fluctuations and long-range temporal dependencies. During training, a specific-energy auxiliary constraint guides the shared representation toward operating patterns associated with energy efficiency. During inference, window-level anomaly scores are obtained from diffusion denoising errors, and the decision threshold is determined from validation-set score quantiles. Using 7476 industrial production records for model development and evaluation, the test-set results show that the detected abnormal windows have 47.83% higher specific energy consumption and 62.17% higher total energy consumption than normal windows, with Cohen’s d values of 1.27 and 1.60, respectively. Compared with PCA, Isolation Forest, autoencoder-based models, an energy-constrained Transformer autoencoder, and diffusion-based baselines, the proposed method produces stronger post hoc energy-consumption differences, while computational cost analysis further clarifies its practical trade-off for offline diagnosis and periodic screening. Group analysis, typical-window diagnosis, and perturbation validation further support its applicability for energy-efficiency diagnosis and high-consumption operating-condition screening in slab reheating furnaces. Full article

Ground-penetrating radar (GPR) is a rapid and non-destructive technique for road sub-surface distress (RSD) detection. However, reliable interpretation of GPR images remains challenging because subsurface anomalies often present weak boundaries, ambiguous textures, and high similarity to non-distress targets. This study proposes a cross-verification intelligent algorithm that exploits complementary information from different views of 3D GPR data. Three YOLO-based detectors are trained on view-specific GPR images to identify RSD-related targets, including voids, loose structures, and manholes. By sequentially verifying detection results across different views, the proposed method improves recognition reliability under vague subsurface imaging conditions. The models are trained and evaluated on an expert-annotated field 3D GPR dataset containing 2134 location-level multi-view samples. At the selected operational thresholds, the complete cross-verification procedure achieved 95.9% precision and 98.6% recall for RSD detection in the testing subset. In a field evaluation on 15 roads, all 69 RSD locations in the expert-identified reference set were matched by automatic indications. When integrated into an automatic detection system, the method reduced manual inspection workloads by approximately 90% while maintaining high field reliability. These results demonstrate the potential of multi-view cross-verification for post-survey RSD screening and expert-assisted review. Full article

As a fundamental technique in geophysical exploration, gravity inversion plays a crucial role in geological structure interpretation and mineral resource assessment. Recent advances in deep learning, particularly the remarkable capabilities of convolutional neural networks (CNNs) in image recognition, object detection, and semantic segmentation, have provided innovative solutions to nonlinear inverse problems in geophysics. However, conventional data-driven approaches often yield physically unrealistic inversion results due to the inherent non-uniqueness of solutions. This study proposes a novel 3D deep-learning gravity inversion framework incorporating well-logging prior constraints, which establishes a density-position relationship-based constraint mechanism through synergistic integration of surface gravity anomaly data and downhole petrophysical parameters. We develop a new neural network architecture that incorporates well-logging data as hard constraints for gravity inversion while preserving both density characteristics and spatial information of subsurface formations. Significantly, we implement the Convolutional Block Attention Module (CBAM) during network optimization, enabling selective enhancement of lithological features during backpropagation. This attention-guided mechanism achieves robust coupling between potential field anomalies and petrophysical parameters from well logs, substantially improving the spatial accuracy of 3D density reconstruction. Numerical experiments demonstrate that our method can accurately recover the density distribution of subsurface ore bodies. Comparative results show that the integration of well-logging information significantly enhances both solution reliability and structural consistency when compared to purely surface data-driven approaches. In practical application to field data from the San Nicolas sulfide deposit in Mexico, the proposed method outperforms conventional UNet-based approaches in terms of inversion performance. Full article

Background/Objectives: Branchio-oto syndrome (BOS) is an autosomal dominant disorder primarily associated with pathogenic variants in EYA1, mainly characterized by branchial anomalies, auricular abnormalities, and hearing loss. However, the co-occurrence of inner ear malformations in BOS remains understudied, especially severe malformations. This study aimed to investigate the clinical and genetic characteristics of patients with EYA1-associated BOS, with emphasis on cochlear nerve deficiency (CND). Methods: From January 2020 to April 2026, patients diagnosed with EYA1-associated BOS at an otology outpatient clinic in a tertiary hospital were included. Clinical manifestations, audiological assessments, imaging and genetic findings were analyzed. Results: Thirteen patients (six females and seven males) from eight unrelated families aged 0.3–58.3 years were enrolled. Branchial cleft fistulas and preauricular pits were each observed in 76.9% (10/13) of patients. The mean pure-tone average was 74.3 ± 20.7 dB HL. Eight EYA1 variants (four truncating, two large deletions, and two splicing) were identified. Among these, five were novel (c.320_329del, c.518del, c.1307dupT, c.1475+1G>A, and exon 12–18 deletion). CND was detected in 38.5% (5/13) of patients and 26.9% (7/26) of ears. Patients with CND carried either truncating variants (n = 3) or large deletions (n = 2) of EYA1. No CND was observed in patients with splicing variants. Conclusions: This study identifies five novel EYA1 pathogenic variants and suggests that CND may be a relatively common radiologic feature in EYA1-associated BOS, particularly among patients with truncating variants or large deletions, although larger studies are needed to confirm this association. Full article

Automated visual inspection is a cornerstone of modern manufacturing quality assurance, yet the effectiveness of any detection system is fundamentally bounded by the informativeness of the observations it receives. Most vision–language model (VLM) and reinforcement learning methods for industrial defect detection assume a fixed set of observations and optimize only the reasoning applied to them. We introduce ActiveInspect, which formulates inspection as budget-constrained sequential selection of multi-view, multi-modal evidence. Starting from a pre-acquired observation pool, a single policy selects an additional view or modality, zooms into a candidate region, retrieves a matched normal reference, or terminates with a verdict. The policy is initialized by perception-activated supervised fine-tuning (PA-SFT) and subsequently optimized by group relative policy optimization (GRPO) using inspection-specific rewards. Depth and point-cloud measurements are converted into VLM-compatible geometric renderings, while a structured memory integrates evidence across inspection steps. Evaluation on Real-IAD D³, Real-IAD, MVTec 3D-AD, MVTec-AD, VisA, and MMAD demonstrates a consistent improvement in the accuracy–observation trade-off. On Real-IAD D³, ActiveInspect increases image-level area under the receiver operating characteristic curve (I-AUROC) from 0.890 to 0.906 (mean over three training seeds; $p<0.01$) relative to the passive D³M baseline while reducing the average observation count from 3.0 to 2.7. It reaches 99.8% of the I-AUROC obtained by exhaustive evaluation of all 15 observations while using 18% of that observation count, and it reduces per-sample inference time by a factor of 5.3 relative to the exhaustive scan. The largest gains occur for geometry-dependent defects, including dents, warping, and concavities. Full article

Field vibration monitoring of EMU traction motor bearings is commonly constrained by weak or relative labels, fluctuations in operating conditions, and limited abnormal samples. Under these conditions, learning a normal boundary from a single bearing position may be unstable, and isolated score spikes may lead to unreliable alarms. To address these issues, this study proposes a multi-axle reference and temporal-consistency-enhanced Deep SVDD framework, termed MA-TC-Deep SVDD, for field anomaly detection of EMU traction motor bearings. Unlike closed-set fault diagnosis that requires known fault labels, the proposed framework focuses on identifying deviations from the stable operating regime. First, a compact 10-dimensional time-frequency representation is constructed from valid vibration segments. Second, stable samples from the target bearing position and screened stable samples from other monitored positions on the same EMU are organized as a multi-axle reference set for one-class normal-boundary learning. Third, feature recalibration, temporal-consistency regularization, reference-score standardization, causal smoothing, and consecutive-alarm judgment are incorporated to improve robustness against field disturbances. The anomaly-prior-guided health-state interpretation module is retained only as post hoc evidence for describing severity evolution and does not feed back into the anomaly detection threshold. Field data collected from an in-service EMU over D1–D5 are used for validation. The results show that bearing position 1 has low anomaly scores on D1–D2, exhibits transitional deviation on D3, and shows persistent state deviation on D4–D5, while the other monitored positions remain comparatively stable. Under the current weak-label evaluation protocol, MA-TC-Deep SVDD achieves higher average anomaly detection performance than the compared baseline methods, with AUC = 0.909, AP = 0.872, Precision = 0.887, Recall = 0.802, F1 = 0.843, and FAR = 0.047. These results indicate that the proposed framework can provide field anomaly-warning and severity-oriented interpretation under weak-label monitoring conditions. However, it should not be interpreted as a replacement for disassembly-confirmed fault-type diagnosis or remaining useful life prediction. Full article

The growing need to assess the health of components has made fault detection an increasingly important research topic in recent years. While traditional techniques remain widely used, the expansion of artificial intelligence (AI) has introduced innovative approaches. Among these, autoencoders have demonstrated significant potential for detecting anomalies in electric motors. The aim of this paper is to identify which signals are most suitable for AI-based fault detection in permanent magnet synchronous motors (PMSMs). To achieve this, in addition to analyzing acceleration signals, which are commonly studied in the literature, this work broadens the investigation by including current, voltage, and temperature signals acquired from different positions. The proposed method is tested during endurance tests, where motors operate under highly variable and demanding operating conditions. The collected signals are then analyzed using a 1D convolutional neural network autoencoder (1D CNN AE) to detect possible faults. The results highlight the importance of considering not only acceleration but also alternative monitoring signals. In particular, temperature measurements proved to be crucial for identifying and localizing specific faults, while current and voltage provided valuable insights into motor behavior, such as changes in control strategy, even when they are not directly correlated with fault occurrence. Full article

In semiconductor and optoelectronics manufacturing, the reliability of turbo molecular pumps (TMPs) is vital for maintaining vacuum integrity and ensuring product yield. However, acoustic monitoring in cleanrooms faces severe challenges due to ambient noise levels routinely exceeding 80 dBA and the scarcity of labeled anomaly data. This study proposes an unsupervised acoustic anomaly detection and localization system to address these issues. The performance of the proposed framework is evaluated using an acoustic dataset collected from operational turbopumps in an industrial semiconductor cleanroom environment, encompassing both normal operations and naturally occurring failure conditions. We introduce a Convolutional Autoencoder (CAE) based on ResNet-18, integrated with a Convolutional Block Attention Module (CBAM) to adaptively suppress high-decibel environmental noise. To enhance sensitivity to structural spectral defects, a hybrid loss function combining Mean Squared Error (MSE) and structural similarity index measure (SSIM) is implemented. Experimental results, supported by rigorous hyperparameter sensitivity analysis, demonstrate that the proposed model achieves an outstanding AUC of 0.9535 and a fault recall of 99.06% under a validation-calibrated threshold, significantly outperforming standard U-Net architectures. Furthermore, the system generates anomaly heatmaps for precise time–frequency localization, enabling explainable diagnostics. With a model inference throughput of 330.28 FPS and an end-to-end processing rate of ≈73× in real time, the proposed framework provides an efficient and robust solution for real-time predictive maintenance in noisy industrial settings. Full article

Real-time video anomaly detection (VAD) under realistic edge constraints—sub-second latency, ≤25 W power, no cloud dependency, and human-interpretable output—remains an open problem. Existing lightweight video convolutional neural networks (X3D, MoViNets) are bound to closed-set training distributions, while recent vision–language-model-based VAD methods (LAVAD, VERA, Holmes-VAD) achieve 80–89% area under the curve (AUC) but rely on datacenter-grade GPUs and Chain-of-Thought (CoT) reasoning that pushes per-segment latency well above one second. This paper reframes the design target from peak accuracy to practical edge deployability and contributes two tightly coupled designs: (i) an edge-optimized inference stack that compresses Qwen3-VL-2B with 4-bit Activation-aware Weight Quantization (INT4 AWQ) and serves it through a TensorRT-LLM C++ runtime on NVIDIA Jetson Orin NX (16 GB, 25 W); and (ii) a fully automatic, CoT-free verbalized prompt optimization in which an 8B optimizer iteratively refines a natural-language definition block $D_t$ using class-balanced (stratified) development batches on a disjoint development subset, with no human editing and no runtime cost on the edge device. Three findings support this framing: (a) the inference stack reduces per-segment latency to 0.25 s, a 7.4× speed-up and 55% memory reduction over a Python/PyTorch baseline; (b) verbalized prompt optimization improves zero-shot AUC from 71.82% (manual prompt) to 76.39%, outperforming GPT-4- and Gemini-Pro-generated prompts (74.12% and 74.35%) under the same edge backbone; and (c) single-frame input attains the highest mean AUC among one-, five-, and eight-frame windows—statistically comparable to the five-frame setting—while offering the lowest latency, making it the preferred operating point under the edge budget. While the absolute AUC (76.39%) is below recent server-side methods (CLIP-TSA 87.58%, VadCLIP 88.02%, Holmes-VAD 89.51%), our framework is the only one in this comparison that operates entirely on a ≤25 W edge device, providing a deployment-oriented operating point on the accuracy–feasibility frontier of VLM-based VAD. Full article

Accurate localization of concealed fire sources is the fundamental prerequisite for effective coal fire control and mitigation. The self-potential method, as a passive, cost-effective geophysical technique for large-area surveys, exhibits unique advantages in coal fire detection, yet 3D inversion of self-potential data for high-precision coal fire source localization remains insufficiently studied. In this paper, a 3D iterative compact inversion algorithm based on the minimum support functional is adopted to invert self-potential data for 3D localization of underground concealed coal fire sources. The algorithm’s reliability and robustness are verified systematically via numerical simulations at both laboratory and field scales, followed by sandbox experiments with artificial battery sources and burning coal specimens, and finally validated its engineering applicability in the Sandaoba Coal Fire Area in Xinjiang. Numerical results show that the algorithm achieves high-precision inversion of source current density across multiple orders of magnitude, with a relative error below 10%. Sandbox experiments confirm that coal combustion generates a typical dipole self-potential field with distinct negative surface anomalies, whose amplitude is positively correlated with combustion intensity. Field inversion results successfully delineate the 3D boundary of the concealed fire source (buried at 40–140 m), which is highly consistent with borehole temperature measurement data. This study provides a robust technical framework for 3D detection of concealed coal fires, offering strong technical support for precise coal fire governance and hazard mitigation. Full article

The widespread application of drones in complex environments imposes higher demands on flight safety. However, traditional anomaly detection methods often rely on mathematical models or large amounts of labeled samples, which makes them ill suited to address practical challenges such as unmanned aerial vehicle systems, which exhibit strong nonlinear characteristics, a scarcity of fault samples, and highly variable operating conditions. To overcome the above difficulties, this work puts forward a one-dimensional selective kernel convolutional autoencoder (1-D SKCAE) based on Bayesian optimization for unsupervised drone anomaly detection. Relying solely on normal operation data, this model achieves accurate anomaly identification by leveraging the sudden changes in reconstruction error. For the model architecture, this paper designs a multi-scale selective kernel convolution module and combines an attention mechanism to achieve adaptive feature weighting for various receptive fields. This method effectively improves the model’s ability to represent complex operational conditions and subtle fault characteristics. Simultaneously, Bayesian optimization is embedded into the model training process as a hyperparameter search strategy, enabling the adaptive configuration of key hyperparameters to further enhance detection performance. Extensive experiments were conducted using the RflyMAD simulation dataset and the 3DR Solo real flight dataset. The results demonstrate that the 1-D SKCAE outperforms multiple comparative models, exhibiting superior robustness particularly in complex scenarios such as mixed multi-fault superposition. This method enables drone anomaly detection without fault labels, showcasing strong potential for engineering applications. Full article