Search Results (2,543)

Background: Multimodal large language models (MLLMs) show promise in medical imaging, but their performance is highly dependent on prompt engineering. This study systematically evaluates how different prompting strategies affect diagnostic accuracy in clinical laboratory image interpretation. Methods: We evaluated five MLLMs (ChatGPT-4o, Gemini 2.0 Flash, Claude 3.5 Sonnet, Grok-2, and Perplexity Pro (Claude 3.5 Sonnet)) using 177 proficiency testing images across three domains: blood smears (n = 78), urinalysis (n = 50), and parasitology (n = 49). Three prompting approaches were compared: (1) complex multi-choice prompts with 20 diagnostic options, (2) zero-shot open-ended prompts, and (3) two-step descriptive-reasoning prompts. Images were sourced from the Taiwan Society of Laboratory Medicine external quality assurance archives with expert consensus diagnoses. Results: Zero-shot prompting significantly outperformed complex multi-choice prompts across all models and domains (p < 0.001). With zero-shot prompts, Gemini achieved 78.5% overall accuracy (urinalysis: 92.0%; parasitology: 75.5%; blood smears: 64.1%), representing a 17% improvement over complex prompts. Two-step descriptive-reasoning prompts further improved blood smear accuracy by 8–12% for top-performing models, but showed minimal benefit in urinalysis and parasitology. The re-query mechanism (“please reconsider”) improved urinalysis accuracy by 7.6% but had a negligible effect on blood smears and parasitology. Conclusions: Prompting strategy critically determines MLLM diagnostic performance. Zero-shot approaches with minimal constraints consistently outperform complex multi-choice formats. The remarkable performance of general-purpose models in structured domains like urinalysis (>90% accuracy) demonstrates the considerable progress of multimodal AI. However, complex morphological tasks like blood smear interpretation require either specialized prompting techniques or domain-specific fine-tuning. These findings provide evidence-based guidance for optimizing AI integration in clinical laboratories. Full article

During the preliminary design of flight vehicles, i.e., missiles or guided rockets, propulsion system performance serves as a critical determinant of both maximum range and terminal velocity. However, complex grain configurations in solid rocket motors (SRMs) typically require geometric modeling software to obtain burning surface area, which severely constrains efficiency. To address this challenge, this study presents a neural network-enhanced rapid performance prediction and matching optimization framework for solid rocket motors (NN-SRM). In NN-SRM, neural networks are employed to simulate the evolution of key parameters during grain combustion, including burning surface area, grain volume, and moment of inertia. The zero-dimensional internal ballistics equations coupled with one-dimensional steady isentropic flow relations are incorporated into the framework to rapidly obtain thrust curves. A discrete–continuous mixed differential evolution algorithm is further employed to identify the optimal grain configuration that satisfies specific thrust requirements. Results demonstrate that, as for cylindrical, star, and finocyl grains, the neural network achieves R² exceeding 0.95. Finally, thrust matching optimization is conducted on three grains and achieves promising thrust solutions for the conditions of large thrust with short time and small thrust with long time, which demonstrates the effectiveness and practicality of the constructed NN-SRM. Full article

With the high penetration of distributed photovoltaics (PV) in distribution areas, transformer capacity limits and source–load fluctuations have become key factors constraining PV accommodation. To accurately assess the PV hosting capacity under energy storage regulation, this paper proposes an assessment method based on operating region analysis. First, a coordinated operation model for the distribution area is established, incorporating the transformer capacity, energy storage constraints, and power balance. On this basis, the calculation boundaries for the PV hosting capacity are discussed in two scenarios: Model 1 ignores power curve uncertainty, characterizing the geometry of the conventional operating region to find the maximum deterministic hosting capacity ($S_1$) that keeps the region non-empty. Model 2 introduces box-type uncertainty sets for the source and load, proposes the concept of a “Self-Balanced Operating Region”, and constructs a robust feasibility determination model ($f_3$) based on a Min–Max–Min structure. To solve this multi-layer nested non-convex model, an iterative algorithm based on duality theory and Benders decomposition is employed to determine the robust hosting capacity under uncertainty ($S_2$) at the critical point where $f_3$ shifts from zero to non-zero. Case studies show that source–load uncertainty leads to a significant contraction of the operating region, and the robust hosting capacity under uncertainty requirements is strictly less than the deterministic hosting capacity ($S_1>S_2$). This method quantifies the reduction effect of uncertainty on the accommodation capability, providing a theoretical basis for planning high-renewable penetration distribution areas and energy storage configuration. Full article

High renewable penetration reduces effective inertia and increases frequency variability in microgrids, thereby limiting the performance of purely reactive frequency regulation. This paper presents a two-timescale frequency-support strategy based on bidirectional electric vehicles. The main novelty lies in introducing a learning-assisted correction layer between forecast-based aggregate regulation and final EV-level dispatch. Rather than replacing the predictive controller with an end-to-end data-driven policy, this layer uses measured fleet-state information to correct the supervisory aggregate request online before a final feasibility-preserving dispatch stage converts it into executable vehicle-level commands under concurrent power, energy, plug-in, and departure constraints. A supervisory predictive layer determines the aggregate support action from forecasted photovoltaic and load disturbances, whereas a lower real-time dispatch layer redistributes that action across the available fleet. Feasibility is enforced through an explicit projection stage prior to actuation. The method is assessed in simulation using measured campus operating profiles of irradiance, temperature, demand, frequency, and electric-vehicle availability. Across four representative operating days, the proposed strategy reduced the mean cumulative frequency deviation by 30.3% relative to droop control and by 24.7% relative to predictive-only operation, while reducing the mean time outside the admissible frequency band by 22.2% and 20.0%, respectively. Zero post-projection constraint violations were observed in all evaluated cases. These gains were obtained at the expense of higher actuation usage, thereby making the regulation–usage trade-off explicit. Full article

Hydrogen production is gaining increasing importance as a key component of the transition toward carbon-neutral energy systems. In this study, the prediction of hydrogen generation in membraneless alkaline water electrolyzers (MAWEs) is investigated using deep-learning-based time-series modeling. A single-input modeling framework is adopted, where only the system current is used as the input variable. Experimental current signals obtained from long-duration tests conducted at electrolyte concentrations between 5 and 35 g KOH (7200 s per experiment) are employed as the model inputs, while mass-based hydrogen production (in grams) is used as the output variable. Two recurrent neural network architectures, namely Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), are implemented, and their predictive performance is comparatively evaluated using RMSE, MAE, and R² metrics. In addition to deep learning models, classical approaches including Linear Regression, ARIMA, and Naïve Forecast are also considered for comparison. The results show that both models are capable of accurately reproducing the hydrogen-production dynamics across the entire concentration range. In particular, the prediction accuracy improves notably at medium and high electrolyte concentrations, where the coefficient of determination (R²) approaches 0.98. The residual distributions remain narrow and symmetric around zero, indicating the absence of systematic estimation bias. The results also show that classical models can achieve comparable performance under stable operating conditions, while deep learning models provide advantages in capturing nonlinear and dynamic behavior. While LSTM and GRU exhibit comparable accuracy, each architecture provides complementary advantages under different operating conditions. These findings indicate that deep-learning-based time-series modeling constitutes a lightweight and reliable framework for prediction and control applications in MAWE systems. Overall, this study demonstrates the applicability of data-driven models for the dynamic characterization of membraneless water electrolysis. Full article

In this article, we show that the well-known leading low-temperature correction to the Heisenberg–Euler Lagrangian in a constant electromagnetic field arising at two loops can be efficiently extracted from its one-loop zero-temperature analogue. Resorting to the real-time formalism of equilibrium quantum field theory that explicitly separates out the zero-temperature contribution from the finite-temperature corrections, the determination becomes essentially trivial. In essence, it only requires taking derivatives of the Heisenberg–Euler Lagrangian at one loop and zero temperature for the field strength. As a bonus, we then effectively dress the low-temperature contribution at two loops by one-particle reducible tadpole structures. This generates a subset of higher-loop contributions to the Heisenberg–Euler Lagrangian in the limit of low temperatures. We extract their leading strong-field behavior at a given loop order, and finally resum these to all loop orders. Full article

In the open-winding motor fed by a common DC bus, unbalanced inverter common-mode voltage (CMV), zero-sequence components of the permanent magnet flux linkage, and the PWM dead-time effect can induce a zero-sequence current (ZSC) through the inherent current path. For an open-winding five-phase permanent magnet synchronous motor (OW-FPPMSM) applied in an aerospace rocket starter-generator system, two ZSC suppression strategies based on zero-sequence voltage (ZSV) generation mechanisms are proposed in this paper, which improve motor performance in a simple and efficient manner. In the first strategy, the conventional method is modified to enable asynchronous operation of the two inverters, thereby generating the required ZSV pulses. The switching order and time offset between the two inverters are determined by the reference ZSV. The second strategy employs basic voltage vectors with larger magnitudes, resulting in higher DC bus voltage utilization. By adjusting the switching sequence of the second inverter, the ZSC components at the carrier frequency are eliminated. Both strategies also achieve the injection of the third-harmonic current. Finally, the two strategies are further analyzed in terms of the modulation index and ZSV modulation range. Simulation and experimental results verify the effectiveness of the ZSC suppression strategies. Full article

The aim of this study was to investigate the ability of sewage sludge-derived biochar to remove PO₄-P from real biologically treated wastewater. Biochar was produced via the pyrolysis of anaerobically digested sewage sludge pretreated with nanoscale zero-valent iron (nZVI) at concentrations of 3%, 1.5%, and 0.5% (w/w, based on total solids). A sample without nZVI addition was used as a control. The properties of biochar samples were analyzed, including elemental composition, specific surface area, and pore size. PO₄-P removal was evaluated using both batch adsorption and column experiments. The highest adsorption capacity determined in the batch experiment was 2.5 mg/g. When wastewater was passed through columns packed with 0.3–0.6 mm biochar particles at a hydraulic loading rate of 1 m/h, a 3-fold-higher phosphorus retention capacity was obtained in the range of 7.26–7.82 mg/g. The column containing biochar derived from sewage sludge with 3% nZVI accumulated 7% more PO₄-P than the biochar without nZVI. All columns effectively removed phosphates from wastewater (efficiency > 80%) due to the chemical composition of biochar, which mainly contained Fe and Ca elements. In contrast to the batch experiment, the columns were subject to the biological sorption of phosphates via microorganisms, physical retention between particles, and the formation of precipitates on the surface of a column. Full article

Eggshell thickness is a critical indicator for evaluating egg breakage resistance and hatchability, yet traditional measurement methods remain destructive and inefficient. To address this, this study proposes a robust prediction approach by integrating Gradient Boosting Decision Tree (GBDT) feature optimization with an improved CatBoost algorithm. First, a joint strategy of Standard Normal Variate (SNV) and Multiplicative Scatter Correction (MSC) was employed to eliminate spectral scattering noise and enhance organic matrix fingerprint information. Subsequently, GBDT was introduced for nonlinear feature evaluation to adaptively screen the top 50 wavelengths, effectively mitigating the “curse of dimensionality” and multicollinearity in full-spectrum data. A CatBoost regression model was then constructed using an Ordered Boosting mechanism, supported by a dual anti-overfitting strategy that merged 10-fold nested cross-validation with Bootstrap resampling. Experimental results demonstrate that this method significantly outperforms traditional algorithms in both prediction accuracy and generalization. The coefficients of determination (R²) for the calibration and prediction sets reached 0.930 and 0.918, respectively, with a root mean square error of prediction (RMSEP) of 0.008 mm. Residual analysis confirms that prediction errors follow a zero-mean Gaussian distribution, indicating that systematic bias was effectively eliminated. This research provides a reliable theoretical foundation and technical support for the intelligent grading of poultry egg quality. Full article

Nuclear energy has emerged as a crucial technological solution for ensuring energy security and achieving carbon neutrality goals, given its ultra-high energy density and near-zero carbon emissions against the backdrop of rapid socioeconomic development, increasing energy demands, and accelerated global transition toward low-carbon energy structures. As the core component for energy conversion in nuclear reactors, fuel elements critically determine reactor efficiency and safety performance, with the fission product retention capability of silicon carbide layers in multilayer-coated fuel particles having been thoroughly validated through high-temperature gas-cooled reactor irradiation tests. The precise sphericity control of large-sized UO₂ fuel kernels represents a fundamental requirement for enhancing tristructural isotropic (TRISO) fuel particle performance and advancing Generation IV nuclear power plant development. This study presents a sphericity control strategy based on sol–gel processing that synergistically integrates physicochemical regulation of gelling media with multi-field washing flow field optimization. By implementing silicone oil-mediated interfacial tension gradient control, we effectively suppressed gel sphere destabilization while developing an innovative three-phase sequential washing technique involving kerosene washing, anhydrous ethanol interfacial transition, and ammonia solution replacement, which significantly enhanced mass transfer diffusion in stagnant liquid films and revolutionized fuel microsphere washing technology with improved efficiency and quality. Experimental results demonstrate that this integrated approach increases kernel sphericity qualification to 99.8%, reduces washing solution consumption by 79%, and achieves an average sphericity of 1.03. The research establishes a coupling mechanism between gelling media and multi-field washing processes, elucidating the synergistic effect between interfacial tension regulation and washing optimization, thereby providing both theoretical foundations and engineering application basis for the precision manufacturing of high-performance nuclear fuels. Full article

Background: Single-target Alzheimer’s disease (AD) therapies have repeatedly failed to modify disease progression, highlighting a critical mismatch between multifactorial pathology and reductionist pharmacology. Methods: We developed a representation learning framework using Knowledge-guided Pre-trained Graph Transformers (KPGT) to enable rational multi-target drug discovery, analyzing 2446 molecules across APP, PSEN1, and VCP. Results: KPGT captured target-specific mechanistic signatures with 99.35% classification accuracy. Geometric midpoint analysis identified 15 bridging candidates with mean pIC₅₀ 8.09. We discovered two orthogonal molecular feature signatures, structural features driving multi-target breadth versus chemical features determining single-target potency, with zero descriptor overlap. Chemical orthogonality (d = 3.86) outperformed functional similarity for predicting synergistic pairs, with 95% overlap between multi-target molecules and synergistic combinations. Conclusions: This framework operationalizes systems-level AD drug discovery through interpretable representation learning. Full article

Background/Objectives: This paper investigates the application of radial basis function (RBF) interpolation to adapt the Toyota Human Model for Safety (THUMS) version 6 finite element (FE) models to diverse anthropometric profiles using ANSUR II data. The research focuses on generating personalized human body models (HBMs) across 50th, 80th, and 98th percentiles for both sexes in standing and seated postures, evaluating mesh quality with quantitative metrics, and assessing posture-dependent transformations. Methods: The geometric accuracy for the standing configuration was quantified using DICE similarity coefficients and the 95th percentile Hausdorff distance (HD95). Results: While global whole-body DICE similarity averaged approximately 0.40 due to an inherent variability in distal limb positioning, regional analysis demonstrated strong volumetric overlap in the critical chest and torso regions with DICE values ranging from 0.80 to 0.88. Regional HD95 values were within 20–30 mm across most of the surface area. Surfaces distance analyses showed that more than 95% of the nodes were within $\pm 20$ mm of the target surfaces with the distribution centered near zero across all the percentiles. The mesh quality for both standing and seated morphs demonstrated low violation rates with the aspect ratio being 28% to 30%, while warpage, skewness and, Jacobian determinants were less than 15%. The seated morphs preserved anatomical alignment and posture despite mesh density differences between the postures. Conclusions: These findings indicate that the morphing process preserves anatomical fidelity while highlighting the need for further optimization to mitigate localized distortions in dynamic simulations. Full article

The expansion and ongoing refinement of control solutions for three-phase microgrids are key enablers in the transition from conventional distribution networks to smart microgrids. By integrating distributed generation, a microgrid can operate in either grid-connected or island mode. One of the major technical challenges in microgrid operation is mitigating or eliminating phase power unbalances. Unbalanced single-phase loads, combined with unbalanced and intermittent single-phase generation, can produce adverse effects on both energy efficiency and power quality. Unlike conventional distribution networks, microgrids may exhibit bidirectional power flows, which can occur simultaneously on all phases or differ from phase to phase. This paper introduces new analytical expressions for sizing a balancing capacitive compensator (BCC) for three-phase four-wire systems and derives a simplified sizing algorithm. The approach is validated through a numerical study using a Matlab/Simulink model of a low-voltage three-phase microgrid with high penetration of single-phase loads and single-phase distributed sources. The BCC is installed at the point of common coupling (PCC) between the microgrid and the main grid. Three operating regimes (cases) of the microgrid were analyzed, considering three compensation scenarios (sub-cases) for each: 1—without compensation, 2—with balanced capacitive compensation (classical), and 3—with unbalanced capacitive compensation (with BCC). For each of the three regimes (cases), the use of the BCC determines, at the PCC, in addition to the cancellation of the reactive component of the positive sequence current, the cancellation of the negative- and zero-sequence currents. In other words, the BCC–microgrid assembly is seen from the main grid either as a perfectly balanced active power load or as a perfectly balanced active power source. Thus, the BCC prevents the propagation of the unbalance disturbance in the main grid; in the considered case study, this also results from the cancellation of the negative- and zero-sequence components of the phase voltages measured at the PCC. The results show that the load-balancing capability of the BCC can be extended to power-flow balancing in any network section, including cases where the phase power directions differ. Implemented as a BCC-type SVC or as an automatically adjustable variant (ABCC), the proposed unbalanced shunt capacitive compensation method is effective for mitigating or eliminating bidirectional phase power-flow unbalances. Full article

As zero-energy buildings (ZEBs) become increasingly mandatory, photovoltaic (PV) systems play a key role in increasing on-site energy generation. For staged ZEB certification based on the energy self-sufficiency ratio (ESSR), it is essential to determine the required power generation and to design PV systems with appropriate installation area and location. This study proposes a systematic design decision-support process for PV system planning that links required energy generation to panel installation strategies. The process enables the determination of a feasible installation area and location of PV panels and was implemented as a design-support program. The proposed process was applied to an apartment building under construction with a ZEB certification grade 5. Compared to the existing design, the optimal design reduced the required PV system capacity by 1.7% while increasing the predicted power generation by approximately 2.8%. The reported improvement in energy generation represents a relative comparison between design alternatives evaluated under identical modeling assumptions and therefore remains valid for comparative design decision-making. Field measurements conducted at a residential building with installed PV systems showed that the predicted power generation is consistent with measured trends, supporting comparative design evaluation and feasibility screening in early-stage PV planning. The developed design process provides a practical framework for early-stage PV system planning, supporting informed design decisions to meet target energy self-sufficiency requirements in ZEBs. Full article

Train track occupancy detection is essential for railway operation safety and dispatching, yet GNSS-based positioning and track matching can degrade or fail in turnouts and station yards due to multipath, interference, and dense track layouts. This paper presents an IMU-only method to discriminate track-switching events during turnout passage by exploiting the transient change in heading angular velocity. The Z-axis gyroscope measurement (approximately aligned with the track-plane normal) is used as a heading-rate proxy, and a lightweight indicator is constructed from the difference between a short-window moving average and the full-run mean. The full-run mean further serves as an in situ approximation of the gyroscope zero bias, alleviating the need for pre-calibration and improving robustness to systematic drift. A fixed discrimination threshold is determined from stationary gyroscope noise statistics, and the minimum effective operating speed is derived by combining gyro noise characteristics with the kinematic relationship among train speed, turnout curvature radius, and heading rate. Field experiments conducted from January to April 2025 on three railway sections covering 27 turnouts (300 turnout-passage events) show that, using a constant threshold $T_0=0.002\mathrm{rad}/\mathrm{s}$, the proposed method achieves 100% track-switching discrimination accuracy within 5–40 km/h, without requiring track maps, GNSS, or prior databases. Full article