187037 research outputs found
Sort by
METHOD, COMPUTER DEVICE, AND COMPUTER PROGRAM FOR Deep-Learning-Based Precipitation Nowcasting with Ground Weather Station Data and Radar Data
No full text지상관측 자료와 레이더 반사도를 활용한 심층학습 기반 초단기 강수 예측을 위한 방법, 컴퓨터 장치, 및 컴퓨터 프로그램이 개시된다. 강수 예측 방법은, 대상 지역의 레이더 관측 자료와 지상 관측 자료를 수집하는 단계; 및 상기 레이더 관측 자료와 상기 지상 관측 자료를 입력으로 하는 딥러닝 기반 강수 예측 모델을 통해 상기 대상 지역에 대한 강수 예측을 수행하는 단계를 포함할 수 있다
Electro-optic sampling of the electric-field operator for ultrabroadband pulses of Gaussian quantum light
No full textQuantum light pulses (QLPs) can be described by spatio-temporal modes, each of which is associated with a quantum state. In the mid-infrared spectral range, electro-optic sampling (EOS) provides a means to characterize quantum fluctuations in the electric field of such light pulses. Here, we present a protocol based on the two-port EOS technique that enables the complete characterization of multimode Gaussian quantum light, demonstrating robustness to both the shot noise and cascading effects. We validate this approach theoretically by reconstructing a multimode squeezed state of light generated in a thin nonlinear crystal driven by a single-cycle pulse. Our findings establish the two-port EOS technique as a versatile tool for characterizing ultrafast multimode quantum light, thereby broadening the reach of quantum state tomography. Potential applications include the characterization of complex quantum structures, such as correlations and entanglement in light and matter. Further, extensions to study multimode non-Gaussian QLPs can be envisaged.
Precipitation sequence of nanoscale precipitates in δ-ferrite of CF8M during short-term thermal aging: Role of Ni and Cu
No full textThe aim of this study was to investigate the precipitation sequence of several thermal aging-induced nanosized precipitates in delta-ferrite of CF8M cast austenitic stainless steels (CASSs). The focus was on the effect of Ni and Cu on precipitation behavior during short-term thermal aging. The CF8M CASSs, with different Cu and Ni concentrations and ferrite content, were subjected to thermal aging at 375 degrees C for 1000 h and at 400 degrees C for up to 1500 h. In high-Cu and low-Ni CF8M, spinodal decomposition along with Cu-rich precipitates were formed after 300 h of aging at 400 degrees C while G-phase and omega-phase precipitates were formed after 650 h of aging at 400 degrees C and after 1000 h of aging at 375 degrees C. However, in low-Cu CF8M only spinodal decomposition and G-phase precipitates at rare locations were observed up to 1000 h of aging at 375 degrees C and 400 degrees C. In low-Cu and high-Ni (low-ferrite) CF8M, spinodal decomposition of delta-ferrite followed by simultaneous formation of both G-phase and omega-phase precipitates was observed after 1500 h of aging at 400 degrees C. However, in low-Cu and low-Ni (high-ferrite) CF8M only spinodal decomposition with G-phase at some localized positions was observed for aging up to 1500 h. The addition of Cu accelerated both spinodal decomposition and precipitation of G-phase and omega-phase in delta-ferrite.
Multi-objective generative design framework and realization for quasi-serial manipulator: Considering kinematic and dynamic performance
No full textThis paper proposes a framework for optimizing the linkage mechanism of a quasi-serial manipulator for target tasks. The process is illustrated through a case study of a two-degree-of-freedom (2-DOF) linkage mechanism, which significantly influences the workspace of the quasi-serial manipulator. First, a diverse set of quasi-serial mechanisms is generated with workspaces that satisfy the target task and is converted into three-dimensional computer-aided design (3D CAD) models. Then, kinematic and dynamic analyses are conducted to compute workspace and payload torque labels for surrogate model training. Adaptive sampling is employed to identify an appropriate amount of training data for accurate prediction, while ensemble models are utilized to enhance robustness. After model training, a multi-objective optimization problem is formulated under the mechanical and dynamic constraints of the manipulator. The design objective is to recommend quasi-serial mechanisms with optimized kinematic (workspace) and dynamic (payload torque) performance that fulfill the target task. To explore the underlying physics of the Pareto solutions obtained via the Non-Dominated Sorting Genetic Algorithm (NSGA-II), various data mining techniques are applied to extract design rules that offer practical guidance. Finally, a detailed manipulator is realized using 3D-printed parts with topology optimization, and its performance is verified through a payload test. These results demonstrate the potential of the proposed framework for broader mechanism applications and its capability to support practical design decisions through design rule extraction.
Dbanet: a dual branch aggregation network for real-time semantic segmentation of omnidirectional images in maritime environments
No full textWe introduce DBANet, a dual-branch aggregation network designed for efficient and real-time semantic segmentation of omnidirectional images in maritime environments. To support research and evaluation in this area, we also present the maritime omnidirectional semantic segmentation dataset, which fills the gap in maritime omnidirectional image segmentation. While omnidirectional vision systems are increasingly popular for their 360-degree perception capabilities, their large field of view imposes significant computational demands, and comprehensive evaluation methods for semantic segmentation in such scenarios remain limited. Our approach addresses these challenges by providing a robust and computationally efficient solution applicable to intelligent perception for maritime surface vehicles. Experimental results highlight the performance of DBANet, achieving 92.36 mIoU at 4.94 FPS on the MODSS dataset and 85.08 mIoU at 30.25 FPS on the MaSTr1325 dataset, outperforming state-of-the-art models in both accuracy and efficiency.
Automated and Programmable Cell-Free Systems for Scalable Synthetic Biology with a Focus on Biofoundry Integration
No full textCell-free protein synthesis (CFPS) has been used as a transformative technology in synthetic biology, providing a programmable, scalable, and automation-compatible platform for biological engineering. Freed from the limitations of cell viability and growth, CFPS enables rapid design iteration, precise control of reaction conditions, and high-throughput experimentation. Recent integration of CFPS with biofoundries-automated, high-throughput biological engineering platforms-has dramatically accelerated the Design-Build-Test-Learn cycle, facilitating applications such as enzyme engineering, metabolic pathway prototyping, biosensor development, and remote biomanufacturing. Advances in automation technologies, including liquid-handling robotics and digital microfluidics, have further enhanced the scalability and reproducibility of CFPS workflows. Additionally, coupling CFPS with machine learning has enabled predictive optimization of genetic constructs and biosynthetic systems. This review highlights the technological innovations driving the convergence of CFPS and automated biofoundries, outlining current capabilities, challenges, and future directions toward programmable, scalable, and distributed biological engineering.
