Search Results (5,456)

The slow evolution and inconspicuous nature of buffer-layer ablation in high-voltage cables pose a significant challenge for early fault diagnosis. To tackle this issue, we propose a hybrid diagnostic approach that integrates frequency-domain impedance measurement with a convolutional neural network (CNN). A cable simulation model is first established using transmission-line theory and a distributed-parameter framework. We examine the impedance and phase responses at the cable’s sending end, revealing a consistent decreasing trend with rising frequency alongside periodic resonant peaks. The simulator generates a diverse set of spectral signatures corresponding to various cable health states. The CNN then extracts discriminative features from these waveforms, and a probabilistic clustering preprocessing step further refines the input data. Experimental results on a test set of 78 samples—comprising 52 experimentally measured normal spectra and 26 experimentally calibrated simulated spectra for mild and severe ablation—demonstrate a classification accuracy of 0.95, confirming that the proposed methodology enables reliable, non-intrusive detection of buffer-layer ablation without cable disassembly. Full article

Methanol is a critical molecular marker for the early aging of oil-paper insulation, and its rapid detection is highly valuable for insulation condition assessment and the fault warning of power transformers. Widely used chromatographic methods require sophisticated pretreatment workflow and are not suitable for in situ monitoring. Non-destructive spectroscopic methods remain challenging due to the intrinsically small cross section of trace molecules in complex liquid environments. The rapid, direct detection of trace methanol in an oil mixture has yet to be demonstrated. In this study, a high-performance Raman-enhancing substrate was developed through hierarchical microstructure regulation, combining microscale light-trapping structures and nanoscale field-confinement sites to sense the weak Raman response of methanol in transformer oil. Direct detection of ppm-level methanol in the oil matrix was achieved, without additional adsorption enrichment or other complicated pretreatment procedures. The characteristic Raman band of methanol in transformer oil was identified, and a quantitative sensing method was established. Furthermore, the intrinsic temperature-dependent Raman response of methanol was investigated to evaluate the stability of its characteristic fingerprint bands over a broad temperature range. This work demonstrates a rapid, sensitive, and pretreatment-free spectroscopic strategy for trace methanol detection in complex oil matrices, and also sheds light on the high-sensitivity detection of small molecular markers in complex liquid environments. Full article

Petrochemical plants are complex facilities composed of interconnected equipment to produce essential products for daily human activities. In view of the adoption of the Industrial Internet of Things (IIoT), these facilities use various sensors, including those for level, flow, temperature, and pressure, to monitor and control operations. Several process faults can be detected and classified using data from these sensors. Detecting and classifying these faults helps mitigate production losses, improve product quality, decrease environmental concerns, and improve human safety. Conventional static learning, which learns from historical data, and continual learning, which learns incremental tasks with known preset boundaries, fall short on asynchronous online unknown streaming boundary tasks such as the OpenWorld Problem, which is the true representation of real-world dynamic plant scenarios. To tackle these challenges, we propose a novel online asynchronous framework termed OD-ViTFScL. Our framework utilizes a dual-attention backbone, which concatenates orthogonal information from attention on the vertical time-series data and horizontal inter-sensor relationships. The fused MLP from the temporal and sensor streams learns independent weights for each orthogonal axis. Our framework also uses ODIN, an out-of-distribution technique, to trigger the arrival of a new fault class in the streaming data. Extensive experiments on the Tennessee Eastman Process (TEP) and Case Western Reserve University (CWRU) bearing datasets validate that our proposed OD-ViTFScL outperforms other state-of-the-art fault detection and diagnosis (FDD) approaches. Full article

The proliferation of inverter-based resources (IBRs) and energy storage systems in power grids has led to a decline in the short-circuit ratio (SCR) and reduced fault current magnitudes, compromising the reliability of conventional overcurrent relays (OCRs) and creating protection blind zones. To resolve these vulnerabilities, this paper proposes an electromagnetic repulsion-based Superconducting Fault Current-Limiting Circuit Breaker (SFCL-CB) utilizing a flux-lock type mechanism. Unlike protection schemes that rely on secondary accessories such as current and potential transformers which introduce computational delays, the proposed SFCL-CB utilizes a driving force that scales with the square of the current gradient ((di/dt)²). This characteristic allows the device to distinguish low-magnitude fault currents from transient load growths. Through duality-based modeling and parametric simulations in PSCAD/EMTDC, the structural and electrical design configurations—including the coil turns (N₁, N₂) and the superconducting quench resistance (R_SC)—were optimized. The simulation results verify that the proposed SFCL-CB substantially reduces the OCR blind zone, securing fault clearance within a maximum of 0.131 s in the regions cleared under low SCR and high-fault-resistance conditions, while achieving mechanical separation in 0.0048 s under robust power system. This SFCL-CB offers an alternative to enhance the protection reliability and operational stability of distribution power system. Full article

The rapid expansion of residential photovoltaics (PV) increases the need for long-term evidence on system performance under real operating conditions. This study compared four grid-connected rooftop PV microinstallations (4.69–5.04 kWp) located in the Sandomierz Basin, southeastern Poland, over 2022–2025. Monthly alternating-current production from SolarEdge monitoring was combined with regional sunshine duration and mean air temperature from the IMGW Sandomierz station, providing 192 installation–month observations. Specific yield, capacity factor, a model-based estimate of the performance ratio (PR_AP), Pearson correlations, ordinary least-squares regression, and sensitivity analysis of a documented failure were applied. Mean annual specific yields were 1077.7, 1061.7, 908.6, and 779.6 kWh/kWp for PV-I–PV-IV, respectively, while mean PR_AP estimates were 82.2%, 82.5%, 70.6%, and 65.5%. Sunshine duration was strongly correlated with monthly specific yield (r = 0.830–0.983; p < 0.001), and the combined model explained 88.8% of its variability. Excluding the zero-output failure month of PV-IV increased R² for the sunshine–yield relationship from 0.689 to 0.812 and improved the combined-model fit from 0.888 to 0.921. Greater nominal capacity did not guarantee higher normalized productivity. Regional solar-resource information should therefore be complemented by monitored operational data to support design, benchmarking, fault detection, and local distributed-energy planning. The findings also support SDG 7 by providing evidence for more reliable, locally adapted planning and operation of household photovoltaic systems. Full article

High-dimensional engineering datasets often combine class imbalance, noisy structures, and limited ground truth, making unsupervised analysis difficult to evaluate reliably. This study quantifies how three properties—partition stability, minority class recovery, and topological interpretability—are traded off across clustering methods, using a capacity-matched 25-seed comparison on a TCGA-derived RNA expression dataset (10,095 samples, 19 cancer types, 13,634 genes). We compare K-means across cluster counts $k\in \{19,\dots ,400\}$, self-organizing maps (SOMs) across lattice sizes from 25 to 625 nodes, consensus K-means, a granularity-matched SOM-Super20 control, and four modern baselines (HDBSCAN, spectral clustering, Gaussian mixtures, and Leiden). At matched prototype budgets, K-means is both more reproducible and substantially better at recovering minority classes than SOMs: at 400 prototypes, K-means achieves pairwise NMI $0.819$ versus $0.621$ for the $20\times 20$ SOM and recovers the smallest cancers 6–$14\times$ more effectively (pancreas effective coverage $0.760$ vs. $0.054$).Crucially, the SOM does not close this gap even when given more prototypes ($0.07$ at 625 nodes), so, under matched capacity, minority recovery is better explained by representational capacity and centroid allocation freedom than by topology preservation. The recovery is not free: increasing k overfragments the partition and lowers the pairwise ARI stability ($0.643\to 0.419$ from $k=20$ to $k=400$), while the NMI remains robust ($\approx 0.82$). The hardest minority, pancreas, is recovered only by high-capacity K-means and by no other method evaluated, including SOMs at any size, consensus K-means, SOM-Super20, HDBSCAN, Gaussian mixtures, spectral clustering, and Leiden. The SOM’s distinct value is therefore not stability or recovery but the interpretable two-dimensional topological visualization that it uniquely provides, including a gradient-organized structure that is reproducible across seeds for kidney (weaker for uterus). No single method optimizes all three properties; the appropriate choice depends on whether a task prioritizes reproducibility, minority recovery, or visual interpretability. Because these conclusions follow from the shape of the data and the allocation behavior of the algorithms rather than from biological semantics, we expect them to transfer to high-dimensional imbalanced engineering data, such as those from fault clustering, condition monitoring, and anomaly detection. Full article

Grain boundaries affect the mechanical and functional properties of crystalline materials by influencing interfacial energy, mobility, segregation, and the accumulation of damage. Among the grain boundaries in the grain boundary network, coincidence site lattice boundaries form a particular type of special grain boundary that is characterized by a higher degree of lattice-site coincidence. The present study examined the mechanisms involved in the formation of grain boundaries in a laser-powder-bed-fused AlSi10Mg alloy which had been subjected to two-pass twist equal-channel angular pressing (twist-ECAP). The microstructural evolution, deformation texture, and local orientation gradients were investigated using electron backscatter diffraction (EBSD). Moreover, atomistic simulations were carried out in order to assess the effect of geometrically necessary boundary (GNB)-like dislocation walls on the retention of planar faults. The EBSD results indicated that twist-ECAP considerably refined the microstructure and produced a strong fiber texture. Most of the Σ3 grain boundary segments detected were found in areas dominated by the ⟨110⟩||ED component, showing that their appearance is strongly dependent on the texture. Also, the atomistic modelling showed that the presence of a GNB-like wall led to the retention of planar-fault configurations and thus resulted in the highest number of atomic environments related to faults and extended dislocation-line lengths. These results show that the formation of Σ3 grain boundary segments in severely deformed LPBF AlSi10Mg is a coupled process which is mainly controlled by macroscopic texture selection and is locally assisted by deformation-boundary evolution. Full article

In recent years, high penetrations of inverter-based resources are posing significant challenges to the medium-voltage networks, in which protection schemes based on high fault currents and unidirectional power flow may not perform as expected. This paper proposes a dynamic fault-detection and relay-coordination scheme for a medium-voltage network with high penetration of grid-forming inverter sources. A detailed 33 kV system model comprising six battery energy storage system (BESS) feeders and a four-distributed-load model was built in DIgSILENT Power Factory and tested under various grid-connected and islanded system conditions using the complete short-circuit method. Four simultaneous fault checks, including sequence component analysis, symmetrical voltage variation, superimposed current with voltage restraint, and current waveform analysis, are used to detect the fault in a specific part of the medium-voltage network. After-fault detection, dynamic pickup scaling and relay blocking are coordinated through IEC 61850 GOOSE and DNP3 so only the closest unblocked relay or relay pair trips. The dynamic pickup settings are adjusted considering the ratio of fault levels in the conventional system versus the inverter-based resources-fed medium-voltage system. Simulation results show successful overcurrent coordination retention even when inverter fault current limitation is set at 1.3 p.u. or lower with 10% generation margin. The proposed scheme allows traditional relays with existing infrastructure to function correctly in fully inverter-dominated medium-voltage systems without any synchronous backup. The novelty is the integration of fault confirmation, pickup scaling and a blocking scheme with retention of an independently operating local backup. Compared to fixed grid-connected settings, the proposed scheme recovers islanded-mode pickup values while maintaining primary–backup grading margin for the relay. Full article

Reliable motor-current-based bearing diagnosis requires evaluation on unseen physical bearings and operating conditions. This study uses the Paderborn University benchmark, acquired from a 425 W permanent-magnet synchronous motor (PMSM) test rig, to evaluate time–frequency deep transfer learning under strict bearing-level grouping. Four representations—continuous wavelet transform (CWT), short-time Fourier transform (STFT), wavelet synchrosqueezed transform (WSST), and Fourier synchrosqueezed transform (FSST)—are combined with pretrained CNN backbones across four binary targets: aged-only A/B and artificial-plus-aged C/D, with mixed-fault bearings excluded/included within each pair. The workflow includes pooled-condition candidate discovery, exploratory Main-split leave-one-operating-condition-out (LOCO) screening, and a retrospective multi-split LOCO audit. The audit contains 288 crossed condition–split–seed evaluations. Because pooled test summaries and Main-split LOCO results informed later stages, these evaluations provide descriptive robustness evidence rather than an independent post-selection test. Target A achieved the highest all-split mean balanced accuracy (0.736 for CWT–EfficientNetB0). The pairs for Targets B and C were near-ties, and the Target D ordering reversed when Main was excluded. Across the eight audited candidates, mean sensitivity ranged from 0.618 to 0.948, whereas specificity ranged from 0.092 to 0.564. Target D combined approximately 0.89 sensitivity with an approximately 0.90 false-alarm rate. Thus, operating condition, fault-class composition, bearing split, and error-cost priorities all affect model interpretation. A matched current-domain baseline audit added 336 evaluations using handcrafted-feature RBF–SVM and Random Forest models and a compact raw-current 1D-CNN. The results show that instability is broader than the TF–CNN pipeline but is not uniform across model families: TF candidates were clearly stronger for Targets A and C, feature-based models were stronger for Target B, and Target D remained mixed and protocol-sensitive. A complementary bearing-level source-group analysis quantified six healthy/fault-source categories across the 37 current features; among the eight features with the largest mean between-group variance fraction, only spectral entropy and dominant power fraction preserved the same mixed-versus-non-mixed contrast direction across all four operating conditions. An additional matched Target D reference held the binary target, physical-bearing split, held-out condition, seed, fourth-order FSST representation, ResNet-50 backbone, and training settings fixed while changing the sensing channel. Across 24 matched runs, vibration showed descriptively higher mean balanced accuracy (0.642 vs. 0.503) and specificity (0.476 vs. 0.146), while sensitivity was slightly lower (0.807 vs. 0.859); the comparison does not establish universal modality superiority. Pooled-condition performance is useful for candidate discovery, but credible condition-generalization claims require explicit separation of exploratory selection and confirmatory testing. The numerical findings are specific to the evaluated PMSM benchmark and do not establish universal performance across electric-machine types. Full article

To address noise interference and feature overlap in multiclass diesel-engine fault-state classification, this study proposes a dual-decoder denoising autoencoder with dual-channel feature fusion. Multi-SNR denoising training pairs are encoded by a shared encoder and reconstructed by a general denoising decoder under reconstruction and cross-SNR latent consistency constraints. With the encoder frozen, a normal-manifold decoder is trained on normal samples to generate a normal-state reference. The difference between the decoder outputs forms a normal-manifold residual that characterizes deviations from normal operation. Latent features are processed by a residual multilayer perceptron, whereas residuals are processed by multi-scale 1D convolution with efficient channel attention; gated fusion combines the two channels for classification. On the 3500-DEFault dataset, reconstruction MAE decreased by 93.75% at 0 dB and 90.86% at 15 dB. Anomaly detection based on normal-manifold residuals achieved an ROC-AUC of 0.9749. The proposed method attained an accuracy of 94.21% and a Macro-F1 of 94.28% on the independent test set, improving Macro-F1 by 3.07 percentage points over the DAE classifier. These results demonstrate the effectiveness of the proposed framework for noise-robust multiclass fault-state classification under the evaluated SNR conditions. Full article

Distributed photovoltaic (PV) systems are increasingly important for renewable-energy transformation, prosumer participation, and low-emission electricity markets. However, failures affecting inverters, DC-side components, connectors, protection devices, monitoring units, and PV modules may reduce generation continuity, increase service burden, and weaken user confidence in distributed generation. Previous PV-failure studies have mainly identified failure modes or ranked them according to maintenance priority, whereas service companies require actionable classes linked with inspection intervals and corrective actions. This study develops and empirically applies a Delphi-based dimension-adjusted fuzzy SIWEC-MABAC decision-support framework for classifying PV installation failures into maintenance action classes. The procedure combines a completed three-round Delphi expert panel, linguistic uncertainty modelling using dimension-adjusted fuzzy sets, SIWEC criterion weighting, and MABAC ranking based on distance from the border approximation area. The empirical SIWEC–MABAC results show that safety/fire impact, downtime duration, and detectability difficulty dominate the service classification. Arc-fault-related DC-side damage and cable insulation degradation are assigned to immediate corrective action, whereas junction-box overheating, inverter hardware failure, melted MC4 connectors, hot-spot formation, and DC circuit-breaker failure require short-term preventive inspection. The validated framework supports service triage and inspection scheduling rather than real-time fault detection. A one-dimensional four-cluster check reproduced the same class membership, and TOPSIS cross-validation showed strong rank agreement with MABAC (Spearman rho = 0.951). Full article

Feature engineering has become a common step in AI-based HVAC forecasting, often involving variables calculated from raw building management system (BMS) measurements, such as temperature differences, setpoint tracking deviations, airflow balance indicators, rolling statistics, and temporal or seasonal descriptors. Accurate short-term forecasting can provide a baseline of expected operation for anomaly and fault detection and can support control optimization and operator decision making. However, real-world deployment is complicated due to differences in BMS sensor availability and data quality, as well as the preprocessing and maintenance burden associated with complex feature sets. The actual contribution of these features to the performance of AI forecasting remains underexplored, particularly for short-term prediction of air handling unit (AHU) operation. This study evaluates the impact of features on short-term AHU forecasting using three deep learning (DL) architectures: Temporal Convolutional Networks (TCNs), Long Short-Term Memory (LSTM) networks, and a hybrid CNN–LSTM model. An actual operational AHU dataset from a BMS was used to predict key operational variables, including supply and extract air temperatures, supply and extract fan operating signals, and supply air temperature setpoint-tracking error. Fan signal balance was additionally evaluated as a derived indicator calculated from the two predicted fan signals. Four input configurations were evaluated: (i) full (74 inputs), containing raw BMS measurements, short-cycle temporal variables, engineered and dynamic features, and annual-calendar information; (ii) no annual calendar (68 inputs), identical to full but excluding annual-calendar variables; (iii) raw + short-cycle temporal (20 inputs); and (iv) raw-only (12 inputs). The models used a 60-min input history to forecast the following 30-min at one-minute resolution. Persistence and Ridge models were included as reference baselines. All models were trained and tested on identical data splits and forecasting horizons to ensure a fair comparison. Each DL experiment was repeated across five independent runs, and performance was evaluated using MAE, RMSE, and R². The TCN showed the strongest overall DL performance. Raw-only achieved the highest mean R² in 11 of 15 architecture–target comparisons using just 12 inputs. The best mean DL R² ranged from 0.916 for the fan signals to 0.993 for extract air temperature. Annual-calendar features improved the TCN results but provided no consistent benefit for the LSTM or CNN–LSTM. Ridge slightly outperformed the best DL configurations for temperature-related targets, reflecting the strong short-term continuity of these signals. These findings show that recent raw BMS measurements contain most of the information needed for accurate 30-min AHU forecasting, while explicit seasonal and engineered features provide limited additional value. The resulting simpler models may in the future be used as forecasting components in predictive control and fault detection systems. However, their control and energy-saving benefits must be tested separately. Full article

Chromite is a strategically critical mineral resource in short supply in China, and detecting the spatial distribution of concealed chromitite orebodies in plateau regions remains challenging. The Yarlung Zangbo Suture Zone in Tibet has a surface environment characterized by high elevations, deeply dissected terrain, and thick Quaternary cover. Remote sensing exploration in this setting is constrained by the spatial–spectral resolution trade-off, while Quaternary cover further obscures surface spectral and structural information. Accordingly, taking the Kubinongyue–Mendangga’ermu area in Zhongba County as the study area, we propose a geologically constrained intelligent exploration framework based on high-resolution hyperspectral (HR-HSI) data that integrates super-resolution reconstruction, deep visual interpretation, and ensemble machine learning. The HySure algorithm is introduced to fuse high-resolution multispectral (HR-MSI) and low-resolution hyperspectral (LR-HSI) data, reconstructing a data cube that preserves both fine spatial topological details and continuous hyperspectral signatures. To overcome the severe class imbalance resulting from extremely sparse lineaments and the interference of topographic artifacts, we developed a Trans-CBAM UNet model—integrating the Convolutional Block Attention Module (CBAM) and Transformer architectures—to extract ore-controlling physical boundaries with high connectivity. On this basis, ensemble learning models such as XGBoost and Random Forest were jointly employed to conduct high-dimensional nonlinear classification of subtle metallogenic indicators, including dunite and serpentinization. Quantitative evaluation demonstrates that the reconstructed data significantly enhance the spatial texture representation of hyperspectral imagery while largely preserving overall spectral fidelity, thereby improving the accuracy of lithological unit classification. The Trans-CBAM UNet model achieved an Area Under the Curve (AUC) exceeding 0.90 for fault prediction, providing robust physical constraints for subsequent intelligent extraction. Using the reconstructed HR-HSI data as the primary input and the extracted fault network as a geological constraint, high- and moderate-prospectivity zones for concealed chromitite were delineated. By combining HR-HSI reconstruction, deep-learning-based structural extraction, and ensemble lithological classification, this study establishes a geologically constrained remote sensing workflow for delineating concealed chromitite prospectivity zones. Full article

The Jinsha River Basin on the eastern margin of the Qinghai–Xizang (Tibetan) Plateau is one of the most landslide-prone regions globally. The Batang reach (from Suwalong Township to Changbo Township) lies in the core of the Jinsha River Suture Zone, characterized by complex geological conditions and frequent landslide disasters. To address the limitations of traditional monitoring methods—including difficulty in full-area coverage, high costs, and low efficiency in mountainous canyon terrain—a systematic study of landslide monitoring was conducted using Sentinel-1A SAR data and Interferometric Synthetic Aperture Radar (InSAR) technology. The results demonstrate that SBAS-InSAR, via short-baseline combinations, achieves a monitoring point density of 697 points/km² (6.28 times that of PS-InSAR), offering significant advantages in mountainous canyon areas with dense vegetation and fragmented rock masses. Using this technical framework, a total of 38 active landslides were identified in the study area, including 5 newly detected rapidly deforming hazards, 17 with river blockage risk, and 10 with the potential to bury buildings. The maximum downslope deformation rate reaches −89.56 mm/yr for the landslide near Suwalong Hydropower Station. Landslides are concentrated within 500 m of faults, in weak rock zones, on steep slopes with gradients greater than 30°, and near road-cutting projects. Temporally, landslide deformation shows an evident correlation with rainfall; approximately 70% of annual cumulative deformation occurs within the rainy season. Engineering activities including hydropower station impoundment appear to be associated with elevated deformation rates of local landslides. This study provides a scientific basis for the safety of major infrastructure corridors (e.g., the Sichuan–Tibet Railway) and regional disaster prevention and mitigation. Full article

The 1605 Qiongshan M7½ earthquake resulted in the submergence of more than 100 km² of coastal land. While the eastern segment of the roughly EW-trending Maniao–Puqian Fault (MPF) has been proposed as one of the seismotectonics that triggered this earthquake, the activity of its western segment, which may extend into the Yangpu free trade port, remains highly disputed due to a lack of evidence. On the basis of comprehensive interpretation from shallow seismic exploration and previous trench excavation, two joint drilling geologic sections were bored along each of the two branches of the western segment with a probing depth of approximately 100–120 m to determine whether it was active. Combined with chronological dating, these sections revealed Quaternary sedments with a thickness of approximately 15–50 m, underlain by the Pliocene Haikou Formation (N₂h). Even the lowest part of the Haikou Formation was not drilled to the faults explained by the geophysical data. The suspected “fault” exposed by the trenches did not extend downward and was not formed by faulting. Consequently, the western MPF poses a relatively low surface-rupture hazard to the Lingao–Chengmai area and its surroundings. Full article