Search Results (7,119)

While deep learning-based visual SLAM (VSLAM) has achieved remarkable localization accuracy, its high computational cost and latency remain critical bottlenecks for real-time deployment. To address these limitations, this paper presents NeuroFusion-SLAM, a novel multi-sensor fusion framework tailored for both efficiency and robustness. By incorporating depthwise separable convolution, the framework cuts down model parameters by approximately 40% and training time by 49% while preserving localization accuracy, thus boosting real-time inference performance and computational efficiency in large-scale environments. Furthermore, a global edge optimization strategy is proposed by integrating sliding window optimization with a factor graph framework, which effectively improves the global consistency of the system. Extensive experiments on the TUM-VI and KITTI-360 datasets demonstrate that our system achieves real-time performance with an average latency of 30.4 ms per frame. It runs 3× faster than ORB-SLAM2 and 4× faster than VINS-Mono, while maintaining good localization accuracy. Full article

Residential air conditioning systems are a major contributor to household electricity consumption in tropical regions, where environmental factors such as climate variability and particulate pollution (PM10) can further increase cooling demand and accelerate equipment degradation. This study proposes an Enhanced Deep Q-Network (Enhanced DQN) for energy-efficient and maintenance-aware control of residential split-type air conditioners under dynamic environmental conditions. The proposed method integrates several stability-oriented reinforcement learning mechanisms, including Double Q-learning, a dueling architecture, prioritized experience replay, multi-step returns, Bayesian-style regularization via Monte Carlo dropout, and entropy-aware exploration. The framework is evaluated through a two-stage process consisting of a diagnostic benchmark on LunarLander-v3 to assess learning stability, followed by a realistic 365-day simulation driven by Thai weather and PM10 data. Compared with a fixed 25 °C baseline, the proposed controller reduced annual electricity consumption from 5116.22 kWh to as low as 4440.03 kWh, corresponding to a saving of 13.22%. The learned policy also exhibited environmentally adaptive behavior under high PM10 conditions, indicating maintenance-aware characteristics. These findings demonstrate that reinforcement learning can provide robust, adaptive, and sustainable control strategies for residential cooling systems in tropical environments. Full article

High-frequency endoscopic ultrasound is crucial for the early diagnosis of gastrointestinal tumors. However, achieving high axial resolution, deep tissue signal-to-noise ratio, and real-time data processing simultaneously remains a significant challenge in hardware implementation. This paper proposes a miniaturized real-time high-frequency imaging system based on the Xilinx Artix-7 FPGA. To overcome attenuation limitations of high-frequency signals, we employ a 4-bit Barker code-encoded excitation scheme coupled with a programmable ±100 V high-voltage transmission circuit. This effectively enhances echo energy without exceeding peak voltage safety thresholds. At the receiver end, the system utilizes a multi-channel analog front end integrated with mixed-signal time-gain compensation technology. Furthermore, to address transmission bottlenecks for massive echo data, we designed a Low-Voltage Differential Signaling (LVDS) interface logic based on dynamic phase calibration, ensuring stable, high-speed data transfer to the host computer via USB 3.0. Experimental results with a 20 MHz transducer demonstrate that the system achieves real-time B-mode imaging at 30 frames per second. Phantom testing revealed an axial resolution of 0.13 mm, enabling clear differentiation of 0.1 mm microstructures. Compared to conventional single-pulse excitation, coded excitation technology improved signal-to-noise ratio (SNR) by approximately 4.5 dB at a depth of 40 mm. These results validate the system’s capability for high-precision deep imaging suitable for clinical endoscopy applications, delivered in a compact, low-power form factor. Full article

Low-rank recovery has emerged as a powerful methodology for the restoration of degraded images. Conventional low-rank recovery techniques, however, predominantly rely on nuclear norm or weighted nuclear norm minimization to separate sparse noise. A significant limitation of this approach is its dependence on full singular value decomposition, which imposes a substantial computational burden, thereby hindering its practical applicability. This paper presents a novel image denoising model integrating the weighted nuclear norm and deep image prior. The weighted nuclear norm is introduced to accurately characterize the global structural properties of images, ensuring the consistency of the overall image structure after denoising. Meanwhile, the deep image prior is employed to effectively capture local details, which helps avoid the blurring of textures and edges often caused by excessive noise removal. The complementary advantages of the two components enable the proposed model to achieve superior performance compared with existing denoising methods. To efficiently compute the proposed model, we design the bilinear factorization method and the alternating direction method of multipliers. Experiments show that the proposed method outperforms mainstream approaches in both restoration accuracy and computational efficiency, exhibiting rapid convergence and robust algorithm stability, thereby demonstrating excellent comprehensive performance. Full article

With the widespread adoption of various types of Unmanned Aerial Vehicles (UAVs), their non-compliant operations pose a severe challenge to public safety, necessitating the urgent identification and detection of UAV targets. However, in complex backgrounds, UAV targets exhibit small-scale dimensions and low contrast, coupled with extremely low signal-to-noise ratios. This forces conventional target detection methods to confront issues such as feature convergence, missed detections, and false alarms. To address these challenges, we propose a Multi-View Feature Fusion Network (MVFF) that achieves precise identification of small, low-contrast UAV targets by leveraging complementary multi-view information. First, we design a collaborative view alignment fusion module. This module employs a cross-map feature fusion attention mechanism to establish pixel-level mapping relationships and perform deep fusion, effectively resolving geometric distortion and semantic overlap caused by imaging angle differences. Furthermore, we introduce a view feature smoothing module that employs displacement operators to construct a lightweight long-range modeling mechanism. This overcomes the limitations of traditional convolutional local receptive fields, effectively eliminating ghosting artifacts and response discontinuities arising from multi-view fusion. Additionally, we developed a small object binary cross-entropy loss function. By incorporating scale-adaptive gain factors and confidence-aware weights, this function enhances the learning capability of edge features in small objects, significantly reducing prediction uncertainty caused by background noise. Comparative experiments conducted on a multi-perspective UAV dataset demonstrate that our approach consistently outperforms existing state-of-the-art methods across multiple performance metrics. Specifically, it achieves a Structure-measure of 91.50% and an F-measure of 85.14%, validating the effectiveness and superiority of the proposed method. Full article

Ultra-deep high-pour-point oil (waxy crude oil) reservoirs under high-temperature and high-pressure conditions are characterized by severe heterogeneity and poor displacement efficiency, with the crude oil exhibiting a pour point of approximately 47 °C. Using the XH block as a representative ultra-deep reservoir, this study systematically examines the displacement mechanisms of CO₂ flooding and CO₂–water-alternating-gas (WAG) flooding. This study aims to elucidate the CO₂–oil interactions between CO₂ and waxy crude oil, to compare oil recovery and CO₂ retention under different injection modes in media with varying permeability and heterogeneity, and to provide experimental support for field-scale development. Slim tube, swelling, and long-core flooding experiments were conducted under reservoir conditions (139 °C, 57 MPa). The phase behavior between CO₂ and crude oil, as well as its impact on oil volume and flow properties, was analyzed. Moreover, continuous CO₂ flooding and WAG flooding were compared in low-permeability and medium–high-permeability cores, and WAG was subsequently applied to a parallel-core system to quantify the effect of interlayer heterogeneity. Results indicate that while CO₂ achieves miscibility with the waxy crude at reservoir pressure, its contribution to swelling and viscosity reduction is moderate compared to light oils; thus, recovery relies primarily on miscible displacement. Compared with continuous CO₂ flooding, WAG effectively delays gas breakthrough and enlarges the swept volume, leading to higher oil recovery and CO₂ storage efficiency. Increasing permeability reduces flow resistance and significantly enhances the oil recovery factor. In strongly heterogeneous systems, dominant flow through high-permeability channels markedly weakens displacement in low-permeability zones, resulting in lower overall recovery and CO₂ retention. These results indicate that properly designed WAG schemes can improve the development performance of heterogeneous waxy oil reservoirs while simultaneously meeting CO₂ storage requirements. Full article

A comparative analysis of the long-term dynamics of phytobenthos on the Black Sea coast from 1964 to 2020 has been conducted. The aim of the work was to assess changes in species composition, quantittive characteristics, and distribution of bottom vegetation under the influence of natural and anthropogenic factors. The research was carried out at three transects using standard hydrobotanical methods and analysis of climatic data. The results revealed significant structural reorganization of the communities: a decrease in the proportion of key brown algae (Ericaria crinita and Gongolaria barbata) by the middle of the observation period with partial recovery by 2020, an overall increase in biomass and species diversity, and increased role of epiphytes and green algae. An expansion of the depth range of the phytal zone and an increase in the presence of the deep-water species Phyllophora crispa were established. The main drivers of the transformation are increased anthropogenic pressure and climate change, which aligns with global trends. The obtained data are important for developing measures to preserve coastal ecosystems and can be used in monitoring the ecological state of the aquatic area. A promising direction for further research is the quantitative assessment of the role of the macrophytobenthos in this area in carbon sequestration. Full article

In the scenario where renewable energy sources (RESs) are connected to the power system (PS) through a flexible high-voltage direct current (HVDC) transmission system, their output becomes highly intermittent and volatile due to meteorological factors like wind direction and speed. This variability poses significant challenges to the real-time power balance and control of the PS. To address the uncertainties in system operation and the challenges of RES consumption, this paper proposes an artificial intelligence (AI) algorithm-driven collaborative optimization strategy for virtual power plants (VPPs) considering RESs transmitted by flexible HVDC. Firstly, a self-attention mechanism and multiple gated structures are integrated into a long short-term memory (LSTM) deep learning model. This enhancement improves the model’s ability to capture multi-timescale characteristics of RESs, increasing forecasting accuracy and robustness. Based on these forecasts, a total cost optimization model for VPP operation is developed, which includes high penalty costs for wind and solar curtailment. By embedding economic constraints that prioritize RESs usage, the model can reduce waste caused by traditional cost-driven scheduling. Additionally, to solve the high-dimensional nonlinear optimization problem in VPP scheduling, an improved population-based incremental learning (PBIL) algorithm is introduced. It incorporates an elite retention strategy and an adaptive mutation operator to boost global search efficiency and convergence speed. Simulations based on an VPP incorporating typical offshore wind and solar RESs transmitted via flexible HVDC demonstrate that the improved LSTM reduces MAPE by 7.14% for wind and 4.27% for PV compared to classical LSTM, and the proposed method achieves the lowest curtailment rates (wind 10.74%, PV 10.23%) and total cost (43,752 RMB), outperforming GA, PSO, and GW by 10–18% in cost reduction. Simulation results show that the proposed strategy enhances RESs consumption while maintaining system economy under flexible HVDC transmission. This work offers theoretical and practical insights for optimizing PS with high RES penetration and supports the low-carbon transition of new-type PS. Full article

Accurate structural analysis is essential for the design and optimization of aircraft wings; however, repeated high-fidelity finite element analysis (FEA) becomes computationally expensive when embedded in iterative design loops. This study presents a machine learning-based surrogate modeling framework for the efficient analysis and optimization of metallic commercial wing structures. A detailed Airbus A320-like wing model was developed and analyzed in ANSYS 2023 R1 under modal, static, and eigenvalue buckling conditions. The general dimensions of the Airbus A320 wing were used only as a reference; the resulting model is a conceptual benchmark rather than a one-to-one geometric replica or a validated digital twin of a specific aircraft wing. Using Latin Hypercube Sampling, 340 high-fidelity samples were generated, with 300 samples used for training and validation and 40 retained as an independent holdout set. The proposed Pyramidal Deep Regression Network (PDRN), a deep learning-based surrogate model whose architecture is automatically tuned using Bayesian Optimization, was benchmarked against Artificial Neural Networks (ANNs), Ensemble Learning, Support Vector Regression (SVR), and Gaussian Process Regression (GPR). On the unseen test set, the PDRN achieved the best overall predictive performance, with RMS errors of 0.8% for mass, 3.1% for the first natural frequency, 11.5% for load factor, and 11.4% for safety factor. To evaluate its practical utility, the trained PDRN was embedded into a PSO-based optimization framework for mass minimization under minimum safety factor, load factor, and first-frequency constraints. The surrogate-guided optimum was verified in ANSYS and remained feasible, yielding a mass of 10,485 kg, a first natural frequency of 1.4142 Hz, a load factor of 1.307, and a safety factor of 1.158. Compared with direct ANSYS in-the-loop optimization, the proposed workflow reached a comparable feasible design with substantially fewer high-fidelity evaluations. These results demonstrate that the PDRN provides an accurate and computationally efficient surrogate for rapid wing analysis and constraint-driven structural optimization. Full article

The derivations presented in this paper suggest an intimate relationship between geometry and the electroweak sector at the Planck scale. A Lorentz-invariant maximally symmetric stochastically perturbed spacetime transformed to spherical coordinates reveals an emergent Schwarzschild metric, entirely a statistical structure of stochastic spacetime. Similarly, the transition from a maximally symmetric universe with a complex $SU\left(2\right)$ scalar doublet $\varphi$, comprising four independent real scalar fields with a zero vacuum expectation value (VEV), to spherical coordinates at the Planck scale reveals the spontaneously broken electroweak (EW) sector. Working in the unitarity gauge, the resulting EW potential can be simultaneously mapped in space at the Planck scale and across the EW sector. In space, the resulting EW potential includes a deep well within the Schwarzschild sphere and a shallow well just outside corresponding to an accretion disk. The same potential mapped in the EW space provides an entire family of possible sombrero hat potentials with fourth-order coupling specific to a point in space. At the minimum points of the potential in space, inside the Schwarzschild sphere and at the accretion disk, the $\lambda$ corresponding to the Standard Model (SM) fourth-order coupling is instead derived as $\frac{\lambda }{5}$. The factor of $\frac{1}{5}$ is a simple consequence of the conservation of the EW VEV and the fact that the SM formulation of the EW potential does not account for situations where the perturbations in $\varphi$ dominate. A more general formulation of the EW potential restores the SM quartic coupling and preserves $\lambda$ in space. An emergent Higgs field inside the Schwarzschild black hole is found to directly relate to the stochastic spacetime fields normalized by the Schwarzschild radius. The corresponding Higgs vacuum has both a ground and excited state and the possibility of both positive and negative vacuum entropy. Finally, the scalar-field VEV degeneracy in EW space of the metastable Higgs vacuum appears instead differentiated in space with possible probability, tunneling, and entropy implications. Full article

Ballast compaction is a critical factor influencing ballast bed condition and the operational safety of heavy-haul railways. However, existing quantitative evaluation methods often suffer from overly idealized simulation models and limitations in signal processing and assessment frameworks. To address these issues, this study proposes a quantitative analysis approach for ballast compaction by integrating non-uniform medium simulation modeling, wavelet–Singular Value Decomposition (SVD) joint denoising, frequency–wavenumber (F-K) migration imaging, and wavelet packet decomposition (WPD)-based feature extraction. Forward simulations were conducted based on the constructed model, and the proposed methodology was validated using 1.5 GHz (gigahertz, 1 GHz = ${10}^9$ Hz) ground penetrating radar (GPR) data acquired from compaction experiments. The results demonstrate that wavelet–SVD joint denoising effectively suppresses deep coherent noise caused by strong reflections from sleepers, significantly enhancing the identification of deep effective signals and ensuring accurate localization and feature extraction of compaction zones. The Geometric Mean of WPD High/Low-Frequency Energy Ratio (GMHLFER) exhibits a strong positive correlation with the degree of compaction. In simulations, as the proportion of compacted material increased from $9\%$ to $21\%$, the GMHLFER rose from 21.555 to 26.581. In field tests, the value increased from 22.012 to 26.012 as compaction severity progressed from slight to severe, demonstrating stable responses across full-gradient compaction conditions and indicating high robustness and sensitivity. The proposed method provides an effective approach for quantitative characterization of ballast compaction in heavy-haul railways, and offers a promising technical pathway for intelligent inspection and condition assessment of railway ballast beds. Full article

Although trenchless technology is widely recognized for its low-carbon potential, existing assessment models often overlook the significant impact of regional geological variations on energy consumption. Based on the EN 15804 standard and the Input–Process–Output (IPO) model, this study establishes a high-resolution carbon emission assessment framework focusing on the “Upfront Carbon” stages (Modules A1–A5) of public works. An empirical study was conducted on a sewage microtunneling project in Hualien, Taiwan, characterized by a deep burial depth of 12 m and challenging gravel formations (SPT N-value > 50). Life Cycle Assessment (LCA) principles were adopted to quantify the carbon footprint and benchmark the results against international guidelines from the UK (PJA) and Japan (JSWA). The Life Cycle Inventory (LCI) reveals a unit emission intensity of 349 kgCO₂e/m, significantly higher than international benchmarks. Critical findings indicate that this discrepancy is primarily driven by environmental variables—specifically, geological resistance and grid emission factors. Crucially, the sensitivity analysis demonstrates that the physical resistance of the hard gravel layer increased machinery energy intensity by 18.7% compared to baseline soil conditions. This study officially defines this phenomenon as the “Geological Premium.” Additionally, carbon efficiency was found to be profoundly influenced by the regional grid emission factor (Taiwan: 0.495 vs. UK: 0.193 kgCO₂/kWh). This research establishes a localized empirical database and validates the necessity of expanding assessment boundaries to include auxiliary works in geologically complex regions. The developed framework provides a scalable solution for optimizing embodied carbon in urban infrastructure, offering policymakers a robust scientific basis for implementing precise “Green Public Procurement” and carbon budgeting strategies. Full article

Medium-deep tight sandstone reservoirs represent a new frontier for hydrocarbon exploration. Great natural gas exploration breakthroughs have been made in the third member of the Lingshui Formation in the Baodao Sag, Qiongdongnan Basin. However, the characteristics of tight sandstone reservoirs and the controlling factors of sweet spots remain poorly understood. Using thin sections, SEM and petrophysical data, this study analyzes reservoir properties and key factors controlling sweet-spot formation, and establishes a pore evolution model. The results show that the reservoirs are dominated by lithic feldspathic quartz sandstones, with feldspar dissolution pores, moldic pores and intergranular pores as major pore types, with average areal porosities of 3.90%, 3.57%, and 1.31%, respectively. Feldspathic quartz sandstones constitute sweet-spot reservoirs. The average porosity is 10.92%, and the average permeability is 6.73 × 10⁻³ μm². Grain size shows a positive correlation with reservoir quality. Compaction provides the basis for reservoir densification, resulting in a porosity loss rate of 22.0–28.0%, with an average of 24.0%. Dissolution is critical for sweet-spot development, forming secondary pore zones at 3800–3950 m and 4100–4400 m, with the dissolution-induced porosity increment ranging from 5.77% to 8.68% and averaging 7.20%. Late carbonate cementation further enhances reservoir densification, corresponding to a porosity loss rate of 5.70–10.9% with an average of 8.28%. This study provides a theoretical basis for sweet-spot evaluation and hydrocarbon exploration in deep-water areas of the South China Sea. Full article

Natural rubber (NR) is essential to the medical, industrial, defense and transportation industries. Alternative rubber crops are needed to supplement increasing rubber demands which cannot be met by the tropical rubber tree, Hevea brasiliensis, and to protect supplies in the event of a rubber tree crop collapse, political strife or a pandemic disrupting global rubber supply chains. Taraxacum kok-saghyz, rubber dandelion, has high-molecular-weight NR, substantial rubber content and the ability to grow in temperate regions. It can also grow hydroponically or aeroponically in controlled environments. This work presents a scenario-based cost analysis of requirements to scale up hydroponic rubber dandelion to replace the 1 million metric tons of imported rubber consumed annually by United States manufacturers. Two scale-up scenarios were considered: a single-level, deep water culture greenhouse and an indoor, ten-level hydroponic vertical farm built in a warehouse. Fuel usage, operating costs, electricity consumption, beneficial insect applications, fertilizers, cooling, and more were included for each case. The costs of operation and construction were compared to the value of products to determine potential annual profit. Sensitivity analyses revealed several scenarios which would drastically improve the economics of the hydroponic facilities. A combination of multiple factors may allow economic feasibility. Hydroponic rubber dandelion production can be profitable on a small scale (up to 15 MT of TNR/year) provided leafy greens and inulin are included as coproducts. The validity of scaling up such a system to 100,000 MT TNR/year to meet 10% of US manufacturing requirements depends heavily on successful research-based gains in TNR concentration and root size, the difference in TNR price between a commodity price and a specialty NR, and upon whether or not tropical rubber tree NR is able to continue to provide a stable source of NR for the US. Full article

Tropical grasses are increasingly present in farming systems in Brazil. However, a national-scale assessment of this practice’s impact on soil health (SH) and soybean yield has been lacking. In this study, we conducted a meta-analysis of 55 studies published until February 2026, comprising field trials run in 33 locations in Brazil, aiming to assess the effects of deep-rooted tropical grasses as preceding crops on biological indicators of SH and soybean yield. Results showed that grasses (Urochloa spp. and Megathyrsus maximus) promote soybean yield by 15%, representing an average increase of 515 kg ha⁻¹ and an additional revenue of US$198 ha⁻¹. The analysis of forage grass species used, management system (single or intercropped), soybean cultivar (growth habit, life cycle, genetic modification), and edaphoclimatic controlling factors revealed positive effects of tropical grasses on soybean yield under all the study conditions and yield ranges. SH indicators also showed sizeable increment, notably the activity of arylsulfatase (+35%) and β-glucosidase (+31%), followed by acid phosphatase activity (+20%), microbial biomass carbon (+24%), and organic carbon (+11%). The results confirmed the beneficial effects of deep-rooted tropical grasses, highlighting their contribution to sustainable intensification in tropical farming systems due to their ability to enhance SH. This, in turn, leads to increased soybean yield under most agronomic and environmental conditions. Full article