Search Results (297)

To address the triple bottlenecks of data scarcity, oversized models, and slow inference that hinder Cantonese automatic speech recognition (ASR) in low-resource and edge-deployment settings, this study proposes a cost-effective Cantonese ASR system based on LoRA fine-tuning and INT8 quantization. First, Whisper-tiny is parameter-efficiently fine-tuned on the Common Voice zh-HK training set using LoRA with rank = 8. Only 1.6% of the original weights are updated, reducing the character error rate (CER) from 49.5% to 11.1%, a performance close to full fine-tuning (10.3%), while cutting the training memory footprint and computational cost by approximately one order of magnitude. Next, the fine-tuned model is compressed into a 60 MB INT8 checkpoint via dynamic quantization in ONNX Runtime. On a MacBook Pro M1 Max CPU, the quantized model achieves an RTF = 0.20 (offline inference 5 × real-time) and 43% lower latency than the FP16 baseline; on an NVIDIA A10 GPU, it reaches RTF = 0.06, meeting the requirements of high-concurrency cloud services. Ablation studies confirm that the LoRA-INT8 configuration offers the best trade-off among accuracy, speed, and model size. Limitations include the absence of spontaneous-speech noise data, extreme-hardware validation, and adaptive LoRA structure optimization. Future work will incorporate large-scale self-supervised pre-training, tone-aware loss functions, AdaLoRA architecture search, and INT4/NPU quantization, and will establish an mJ/char energy–accuracy curve. The ultimate goal is to achieve CER ≤ 8%, RTF < 0.1, and mJ/char < 1 for low-power real-time Cantonese ASR in practical IoT scenarios. Full article

The ice sheet and subglacial geological environment in Antarctica have become the focus of scientific exploration. The development of Antarctic drilling technology will serve as a crucial safeguard for scientific exploration. However, the extremely ultra-low temperatures and intricate geological conditions present substantial obstacles for drilling operations in Antarctica, and the existing drilling fluid technology cannot satisfy the requirements of efficient and safe drilling. To ameliorate the wettability and rheology of ultra-low-temperature drilling fluids, a new amide material (HAS) was prepared using dodecylamine polyoxyethylene ether, azelaic acid, and N-ethylethylenediamine as raw materials. Experiments using infrared spectroscopy, nuclear magnetic hydrogen spectroscopy, and contact angle indicated that the target product was successfully synthesized. Performance evaluation showed that 2% HAS could achieve a yield point of 2.5 Pa for drilling fluid at −55 °C, and it also gave the fluid superior shear-thinning characteristics and a large thixotropic loop area. This indicated that HAS significantly enhanced the rheological properties of the drilling fluid, ensuring that it can carry cuttings and ice debris. In addition, 2% HAS could also increase the colloidal rate from 8% to more than 76% at −55 °C in different base oils. Meanwhile, the colloid rate was maintained above 92.4% when the density was 0.92~0.95 g/cm³. Mechanism studies showed that HAS increased the zeta potential and decreased the particle size of organoclay. At the same time, it changed the organoclay state from a clustered state to a uniformly dispersed state, and the particle size decreased. It was found that HAS formed a weak gel grid structure through interactions between polar groups, such as amide and imino groups with organoclays particles, thus improving the rheology and wettability of drilling fluid. In addition, HAS is an environmentally friendly high-performance material. Full article

The deposition of large-sized cuttings (or blocky cuttings) is a critical risk factor for stuck pipe incidents during the drilling of deep and extended-reach wells. This risk is particularly pronounced in well sections with long borehole trajectories and low drilling fluid return velocities, where it poses a substantial threat to wellbore cleanliness and the safe operation of the drill string. This study utilizes a self-developed visual experimental platform to simulate the deposition evolution of large-sized cuttings (20–40 mm in diameter) in the annulus under various wellbore inclinations and drilling fluid parameters. The stable height, lateral distribution characteristics, and response patterns of the resulting cuttings bed under different conditions were quantitatively characterized. Building upon this, a theoretical contact friction model between the drill string and the cuttings bed was employed to investigate how the bed height influences hook load during tripping and rotary torque during top drive operation. A mapping relationship was established between cuttings bed structural parameters and the resulting additional loads and torques. Results reveal significant interactive effects among drilling fluid velocity, fluid density, drill pipe rotation speed, and wellbore inclination on both cuttings bed development and associated drill string loads. Strong correlations were identified among these parameters. Based on these findings, a stuck pipe early-warning indicator system is proposed using frictional load thresholds, with clearly defined safety limits for cuttings bed height. Recommendations for optimizing cuttings transport parameters through coordinated control of fluid velocity, density, and rotary speed are also provided, offering theoretical support and engineering guidance for borehole cleaning strategies and stuck pipe risk prediction in large cuttings scenarios. Full article

This study proposes considering the effective re–benefication of coal middlings and other such considered waste materials as a way to ensure that clean coal in coal by–products can be extracted and effectively utilized, saving costs and reducing coal waste. To quantify the clean–coal yield and ash reduction that can be achieved by re–beneficiating four typical by–product streams from the Guobei Coal Preparation Plant (6 Mt a⁻¹) were used for the study. Coking–coal middlings, flotation tailings, and pressure–filter cakes from preparation plants still contain 30–60% combustible matter. Re–beneficiation techniques have been considered to recover this often-wasted coal, reduce waste rock disposal, and cut greenhouse–gas emissions per ton of clean coal produced. Representative samples (n = 4) were collected, sample size–classified as (fine coal particles ≤0.5 mm and coarse particles ≥) and subjected to (i) magnetite removal, (ii) laboratory froth flotation (diesel 507 g t⁻¹, sec–octanol 103 g t⁻¹), and (iii) fine and large particle density separation at 1.3–1.8 g cm⁻³ ZnCO₃ media. Clean–coal yield and ash were measured for each stream and the coal’s particle liberation was examined by SEM. Crushing, grinding and liberation equipment and techniques that aid in the treatment of coal and the re–beneficiation of coal middlings and tailings. The key findings recorded during the experiment are as follows: Flotation of <0.5 mm fractions delivered 46.9–58.3% clean–coal yield at 10.3–17.0% ash. Density separation of 0.5–1.0 mm middlings peaked at 1.4–1.5 g cm⁻³, yielding 34.2% clean coal at 15–18.4% ash. Scanning Electron Microscope analysis confirmed partial liberation as results from re–grinding + second flotation which increased yield by a further 8–12%. A calculated theoretical examination of the preliminary cost–benefit analysis indicates ≈36 CNY t⁻¹≈9 million CNY a⁻¹ in saved disposal costs alone. savings in disposal and 0.25 Mt a⁻¹ additional clean coal for the Guobei plant. The research presented in this paper highlights the current work by Anhui University of Science and technology in collaboration with Guobei coal preparation plant and the results therein achieved. Full article

Most astrophysical systems (except for very compact objects such as, e.g., black holes and neutron stars) in our Universe are characterized by shallow gravitational potentials, with dimensionless compactness $|\Phi |\equiv r_s/R\ll 1$, where $r_s$ and R are their Schwarzschild radius and typical size, respectively. While the existence and characteristic scales of such virialized systems depend on gravity, we demonstrate that the value of $|\Phi |$—and thus the non-relativistic nature of most astrophysical objects—arises from microphysical parameters, specifically the fine structure constant and the electron-to-proton mass ratio, and is fundamentally independent of the gravitational constant, G. In fact, the (generally extensive) gravitational potential becomes ‘locally’ intensive at the system boundary; the compactness parameter corresponds to the binding energy (or degeneracy energy, in the case of quantum degeneracy pressure-supported systems) per proton, representing the amount of work that needs to be done in order to allow proton extraction from the system. More generally, extensive properties of gravitating systems depend on G, whereas intensive properties do not. It then follows that peak rms values of large-scale astrophysical velocities and escape velocities associated with naturally formed astrophysical systems are determined by electromagnetic and atomic physics, not by gravitation, and that the compactness, $|\Phi |$, is always set by microphysical scales—even for the most compact objects, such as neutron stars, where $|\Phi |$ is determined by quantities like the pion-to-proton mass ratio. This observation, largely overlooked in the literature, explains why the Universe is not dominated by relativistic, compact objects and connects the relatively low entropy of the observable Universe to underlying basic microphysics. Our results emphasize the central but underappreciated role played by dimensionless microphysical constants in shaping the macroscopic gravitational landscape of the Universe. In particular, we clarify that this independence of the compactness, $|\Phi |$, from G applies specifically to entire, virialized, or degeneracy pressure-supported systems, naturally formed astrophysical systems—such as stars, galaxies, and planets—that have reached equilibrium between self-gravity and microphysical processes. In contrast, arbitrary subsystems (e.g., a piece cut from a planet) do not exhibit this property; well within/outside the gravitating object, the rms velocity is suppressed and G reappears. Finally, we point out that a clear distinction between intensive and extensive astrophysical/cosmological properties could potentially shed new light on the mass hierarchy and the cosmological constant problems; both may be related to the large complexity of our Universe. Full article

This study aimed to describe and assess differences among playing positions, to determine playing position profiles, and to analyze the relationships between the change-of-direction deficit (CODD) percentage and the other anthropometric and performance variables. A total of 98 young futsal players (age: 17 ± 1 years) from the highest national level in Spain were assessed using a cross-sectional design. Anthropometric variables such as height and body mass were recorded. The performance tests included countermovement jumps; horizontal jumps; sprint tests (10 m and 25 m); change-of-direction (COD) tests, including a 10 m test with one COD of 180° (COD180) and a 25 m test with 4 CODs (V-cut); and the percentage CODD. Furthermore, asymmetries were recorded. The group comparisons were considered statistically significant at p ≤ 0.05 and were supported by the effect sizes and mean differences. Significant differences were found among playing positions, showing that pivots and goalkeepers were significantly taller than left- and right-wingers and defenders (p < 0.05, effect size (ES) = −1.42 to 0.72). Goalkeepers were significantly slower than the rest of the positions in COD180 to the left (p < 0.05, ES = 1.32 to 1.89). A very large association was found between the CODDs of 25 m and 25 m (p < 0.001; r = −0.72). These results suggest that pivots and goalkeepers are taller and larger than the other players. However, aside from goalkeepers, no differences in performance variables were observed among the outfield players. In addition, a lower %CODD is associated with a faster COD performance, highlighting its importance in training. Full article

HgCdSe has recently been proposed as a potential alternative material to HgCdTe for fabricating high-performance infrared detectors. This work presents a study on the growth of high-crystalline-quality HgCdSe materials on GaSb (211)B substrates via molecular beam epitaxy and demonstration of the first prototype HgCdSe-based mid-wave infrared detectors. By optimizing the MBE growth parameters, and especially the thermal cleaning process of the GaSb substrate surface prior to epitaxial growth, high-quality HgCdSe material was achieved with a record XRD full width at half maximum of ~65 arcsec. At a temperature of 77 K, the mid-wave infrared HgCdSe n-type material demonstrated a minority carrier lifetime of ~1.19 µs, background electron concentration of ~2.2 × 10¹⁷ cm⁻³, and electron mobility of ~1.6 × 10⁴ cm²/Vs. The fabricated mid-wave infrared HgCdSe photoconductor presented a cut-off wavelength of 4.2 µm, a peak responsivity of ~40 V/W, and a peak detectivity of ~1.2 × 10⁹ cmHz^1/2/W at 77 K. Due to the relatively high background electron concentration, the detector performance is lower than that of state-of-the-art low-doped HgCdTe counterparts. However, these preliminary results indicate the great potential of HgCdSe materials for achieving next-generation IR detectors on large-area substrates with features of lower cost and larger array format size. Full article

The research and development of the next generation of supersonic transports (SSTs) meets economic and environmental problems. An SST encounters critical challenges, including the need for low fuel consumption, low noise, and low gas emissions. Currently, the feasibility of developing SSTs is increasing through the application of cutting-edge technologies, such as composite materials, advanced electric systems, sustainable aviation fuel, and innovative design methodologies. The object of this study was to perform the conceptual design of a 50-seat supersonic transport utilizing general conceptual design methods. In estimating weight and flight performance, statistical formulae were correlated with data from civil supersonic and subsonic jet transports. For wing sizing, carpet plots were created to explore the optimal combination of wing aspect ratio and wing loading. The results suggested that by utilizing advanced technologies, such as the use of a composite material for the structure, the maximum takeoff weight can potentially be reduced while still meeting design requirements. The constraint of climb gradient largely affects the maximum takeoff weight, and it is anticipated that flight performance at low speeds will be improved. Full article

Background/Objective: The approach to the clinical management of metastatic melanoma patients is undergoing a significant transformation. The availability of a large amount of data from medical images has made Artificial Intelligence (AI) applications an innovative and cutting-edge solution that could revolutionize the surveillance and management of these patients. In this study, we develop and validate a machine-learning model based on radiomic data extracted from a computed tomography (CT) analysis of patients with metastatic melanoma (MM). This approach was designed to accurately predict prognosis and identify the potential key factors associated with prognosis. Methods: To achieve this goal, we used radiomic pipelines to extract the quantitative features related to lesion texture, morphology, and intensity from high-quality CT images. We retrospectively collected a cohort of 58 patients with metastatic melanoma, from which a total of 60 CT series were used for model training, and 70 independent CT series were employed for external testing. Model performance was evaluated using metrics such as sensitivity, specificity, and AUC (area under the curve), demonstrating particularly favorable results compared to traditional methods. Results: The model used in this study presented a ROC-AUC curve of 82% in the internal test and, in combination with AI, presented a good predictive ability regarding lesion outcome. Conclusions: Although the cohort size was limited and the data were collected retrospectively from a single institution, the findings provide a promising basis for further validation in larger and more diverse patient populations. This approach could directly support clinical decision-making by providing accurate and personalized prognostic information. Full article

Green hydrogen production from water electrolysis (WE) is one of the most promising technologies to realize a decarbonized future and efficiently utilize intermittent renewable energy. Among the various WE technologies, the emerging anion exchange membrane (AEMWE) technology shows the greatest potential for producing green hydrogen at a competitive price. To achieve this goal, simple methods for the large-scale synthesis of efficient and low-cost electrocatalysts are needed. This paper proposes a very simple and scalable process for the synthesis of nanostructured NiCo- and NiFe-based electrode materials for a zero-gap AEMWE full cell. For the preparation of the cell anode, oxides with different Ni molar fractions (0.50 or 0.85) are synthesized by the sol–gel method, followed by calcination in air at different temperatures (400 or 800 °C). To fabricate the cell cathode, the oxides are reduced in a H₂/Ar atmosphere. Electrochemical testing reveals that phase purity and average crystal size significantly influence cell performance. Highly pure and finely grained electrocatalysts yield higher current densities at lower overpotentials. The best performing membrane electrode assembly exhibits a current density of 1 A cm⁻² at 2.15 V during a steady-state 150 h long stability test with 1 M KOH recirculating through the cell, the lowest series resistance at any cell potential (1.8 or 2.0 V), and the highest current density at the cut-off voltage (2.2 V) both at the beginning (1 A cm⁻²) and end of tests (1.78 A cm⁻²). The presented results pave the way to obtain, via simple and scalable techniques, cost-effective catalysts for the production of green hydrogen aimed at a wider market penetration by AEMWE. Full article

Introduction: Computed tomography-guided biopsies (CTGB) are essential in diagnosing various conditions, particularly in respiratory medicine, with lung cancer being a primary focus. A significant complication associated with CTGB is pneumothorax, which can occur in up to 26% of cases. At Northumbria Healthcare NHS Foundation Trust, a large interventional service collaborates closely with radiologists and respiratory physicians. This study aims to evaluate the incidence of pneumothorax following CTGB. Methods: A retrospective service review was conducted on all lung parenchymal CTGBs performed between April 2011 and July 2023, with approval from the local information governance. Demographic data and clinical outcomes were analyzed using descriptive statistics. Continuous variables are presented as medians with interquartile ranges (IQR), while categorical variables are reported as frequencies and percentages. Results: A total of 1492 CT-guided lung biopsies were analyzed. The median age of patients was 72 years (IQR 10.5), and 50.9% were male. Pneumothorax occurred in 23.8% (n = 355) of cases. Of these, 159 (44.8%) were detected on post-biopsy CT scans. The average number of pleural passes was 1.8 (range 1–4). Among those with pneumothorax, 53.6% had radiologically evident emphysema. The median forced expiratory volume in 1 s (FEV1) was 1.97 L (IQR 1.04). Sixty-seven percent (n = 234) of patients had no pleural contact, and the median lesion size was 26 mm (IQR 24). Seventy-two percent (n = 255) of lesions with pneumothoraces were less than 3 cm deep. Forty-four percent of biopsies were performed using 18 French gauge tru-cut needles. Of the 355 pneumothoraces, 89% (n = 315) were managed conservatively, with 42 requiring pleural intervention (41 small-bore 12 Fr intercostal chest drains and one pleural vent). Symptoms were initially present in 40 cases, and two cases developed symptoms up to 7 days post-procedure. Conclusions: The incidence of pneumothorax is consistent with expected rates, with more occurrences observed in biopsies of smaller lesions lacking pleural contact, lesions with surrounding emphysema, and cases requiring multiple pleural passes. FEV1 does not appear to influence the risk of pneumothorax. Conservative management is generally effective, without significant complications. Full article

In recent years, wireless sensor networks (WSNs) have been employed across various domains, including military services, healthcare, disaster response, industrial automation, and smart infrastructure. Due to the absence of fixed communication infrastructure, WSNs rely on ad hoc connections between sensor nodes to transmit sensed data to target nodes. Within a WSN, a sensor node whose failure partitions the network into disconnected segments is referred to as a critical node or cut vertex. Identifying such nodes is a fundamental step toward ensuring the reliability of WSNs. The critical node detection problem (CNDP) focuses on determining the set of nodes whose removal most significantly affects the network’s connectivity, stability, functionality, robustness, and resilience. CNDP is a significant challenge in network analysis that involves identifying the nodes that have a significant influence on connectivity or centrality measures within a network. However, achieving an optimal solution for the CNDP is often hindered by its time-consuming and computationally intensive nature, especially when dealing with large-scale networks. In response to this challenge, we present a method based on particle swarm optimization (PSO) for the detection of critical nodes. We employ discrete PSO (DPSO) along with the modified position equation (MPE) to effectively solve the CNDP, making it applicable to various k-vertex variations of the problem. We examine the impact of population size on both execution time and result quality. Experimental analysisusing different neighborhood topologies—namely, the star topology and the dynamic topology—was conducted to analyze their impact on solution effectiveness and adaptability to diverse network configurations. We consistently observed better result quality with the dynamic topology compared to the star topology for the same population size, while the star topology exhibited better execution time. Our findings reveal the promising efficacy of the proposed solution in addressing the CNDP, achieving high-quality solutions compared to existing methods. Full article

The wheat stem sawfly (WSS, Cephus cinctus Norton) is a major insect pest of wheat (Triticum aestivum L.) in North America. Few management tactics exist, and quantifying their efficacy is confounded by the difficulty in monitoring infestation at the field scale. Accurate estimates of WSS infestation are cost prohibitive as they rely on comprehensive stem dissection surveys due to the concealed life cycle of the pest. Consolidating the available management tactics into an effective strategy requires inexpensive, spatially explicit estimates of WSS infestation that are compatible with the large field sizes dryland wheat is often sown to. Therefore, we investigated using multitemporal satellite passive remote sensing (RS) to estimate various metrics of WSS infestation collected from field surveys at the subfield scale. To achieve this, we dissected 43,155 individual stems collected from 1158 unique locations across 9 production wheat fields in Montana, USA. The dissected stem samples from each location were then quantified using the following metrics: the proportion of total WSS-infested stems, proportion of stems with more than one node burrowed through (adequate WSS infestations), and proportion of WSS cut stems only. Cloud-free Sentinel 2 images were collected from Google Earth Engine for each field from across the growing season and sparse multiway partial least squares regression was used to produce a model for total WSS infestations, adequate WSS infestations, and WSS cut stems, for each sampled field. Upon comparing the performance of these models, we found that, on average, the metrics for total (R² = 0.57) and adequate WSS infestations (R² = 0.57) were more accurately estimated than WSS cut (R² = 0.34). The results of this study indicate that multitemporal RS can help estimate total and adequate WSS infestations, but more holistic methods of field level sensing should be explored, especially for estimating WSS cutting. Full article

Accurate identification of tomato ripeness and precise detection of picking points is the key to realizing automated picking. Aiming at the problems faced in practical applications, such as low accuracy of tomato ripeness and picking points detection in complex greenhouse environments, which leads to wrong picking, missed picking, and fruit damage by robots, this study proposes the YOLO-TMPPD (Tomato Maturity and Picking Point Detection) model. YOLO-TMPPD is structurally improved and algorithmically optimized based on the YOLOv8 baseline architecture. Firstly, the Depthwise Convolution (DWConv) module is utilized to substitute the C2f module within the backbone network. This substitution not only cuts down the model’s computational load but also simultaneously enhances the detection precision. Secondly, the Content-Aware ReAssembly of FEatures (CARAFE) operator is utilized to enhance the up-sampling operation, enabling precise content-aware processing of tomatoes and picking keypoints to improve accuracy and recall. Finally, the Convolutional Attention Mechanism (CBAM) module is incorporated to enhance the model’s ability to detect tomato-picking key regions in a large field of view in both channel and spatial dimensions. Ablation experiments were conducted to validate the effectiveness of each proposed module (DWConv, CARAFE, CBAM), and the architecture was compared with YOLOv3, v5, v6, v8, v9, and v10. The experimental results reveal that, when juxtaposed with the original network model, the YOLO-TMPPD model brings about remarkable improvements. Specifically, it improves the object detection F1 score by 4.48% and enhances the keypoint detection accuracy by 4.43%. Furthermore, the model’s size is reduced by 8.6%. This study holds substantial theoretical and practical value. In the complex environment of a greenhouse, it contributes significantly to computer-vision-enabled detection of tomato ripening. It can also help robots accurately locate picking points and estimate posture, which is crucial for efficient and precise tomato-picking operations without damage. Full article

Autonomous inflow control devices (AICDs) offer a promising means of delaying early water breakthrough in heavy oil horizontal wells; yet, current design practices remain largely empirical. A three-dimensional, field-calibrated computational fluid dynamics (CFD) model was developed to establish a mechanistic basis that solves the transient Navier–Stokes equations for turbulent two-phase flow via a volume-of-fluid formulation. Pressure-controlled inflow boundaries were tuned to build up data from four Colombian heavy oil producers, enabling a quantitative comparison with production logs. Model predictions deviate by no more than ±14% for oil rate and ±10% for water rate over a 500-day horizon, providing confidence in subsequent scenario analysis. Replacing a slotted liner completion with optimally sized AICDs lowers cumulative water-cut by up to 93%, reduces annular friction losses by 18%, and cuts estimated life cycle CO₂ emissions per stock-tank barrel by 82%. Sensitivity analysis identifies nozzle diameter as the dominant design variable, with a nonlinear interaction between local drawdown pressure and the oil–water viscosity ratio. These findings demonstrate that CFD-guided AICD design can materially extend wells’ economic life while delivering substantial environmental benefits. The validated workflow establishes a low-risk, physics-based path for tailoring AICDs to reservoir conditions before field deployment. Full article