598 papers
This paper proposes an autonomous, multimodal framework that integrates PTZ cameras, drones, wearable IoT sensors, and a Digital Twin with Vision Foundation Models to provide real-time environmental risk intelligence, predictive safety analytics, and automated emergency responses for the smart firework industry.
This paper proposes Wiener-KAN, an on-board real-time compensation system for electrochemical seismometers that combines Wiener-model linearization with a precomputed, lookup-table-based Kolmogorov-Arnold network to effectively mitigate nonlinear frequency drift and sensitivity errors while maintaining ultra-low latency and minimal resource usage on microcontrollers.
This paper proposes a novel spatiotemporal O(1) resonance computing architecture that eliminates chemical batteries in smart factory mobility by utilizing a 64-byte coordinate vector, virtual quantum processing, and evanescent wave coupling to achieve constant-latency control, 92.5% wireless power transfer efficiency, and zero hacking risk.
This study evaluates the impact of varying TiO₂ nanoparticle sizes (13–38 nm) added to a diesel/waste tyre pyrolysis oil blend on engine performance, emissions, and thermodynamic efficiency, revealing that while larger nanoparticles reduce CO and HC emissions, they simultaneously increase fuel consumption and decrease thermal and exergy efficiencies compared to pure diesel.
This paper introduces a diffusion-based generative model that efficiently samples probabilistic global sea states conditioned on historical wind forcing, offering a computationally accelerated and calibrated alternative to traditional spectral wave models for ensemble forecasting and earth system coupling.
This study demonstrates that incorporating hooked-end steel fibres (up to 40 kg/m³) into concrete designed via the Dreux–Gorisse method significantly enhances 28-day compressive strength by 32.7% and density through crack-bridging mechanisms, albeit at the cost of a 75% reduction in workability.
This paper presents a physics-informed deep learning framework that reconstructs high-fidelity neutron computed tomography images from ultra-sparse angular sampling, reducing acquisition time by over 90% while maintaining quantitative structural integrity and enabling operando studies of fast dynamic processes.
This paper presents a systematic framework and a corresponding spreadsheet-based computational tool for the preliminary dimensioning and characterization of internal combustion engines, integrating geometric, thermodynamic, and dynamic analyses to estimate key component dimensions and performance indicators based on limited initial operating data.
This paper demonstrates that while a neural estimator trained on wall-shear-stress data can accurately reconstruct near-wall flow quantities offline, it fails in closed-loop opposition control due to distribution shifts caused by the controller itself, necessitating spectral consistency constraints and closed-loop retraining to achieve stable, drag-reducing performance.
This study demonstrates that combining the Taguchi method with Random Forest Regression effectively optimizes FDM printing parameters for ABS PA-747H, significantly enhancing tensile and flexural mechanical properties while validating the improvements through statistical analysis and fractographic evidence.