187037 research outputs found
Sort by
Dynamic behavior of duplex stainless steel with improved chloride-induced stress corrosion cracking resistance in drop scenarios for dry storage containers
No full textThis study investigated the safety and structural integrity enhancements in dry storage containers (DSCs) achieved using a novel duplex stainless steel with a composition of Fe-19Cr-4Ni-2.5Mo-4.5Mn, referred to as advanced duplex stainless steel for a canister of dry storage (ADCS). This material was developed to improve resistance to chloride-induced stress corrosion cracking. The safety and structural integrity of DSCs utilizing ADCS and conventional austenitic stainless steel (ASS) were comparatively evaluated using finite element analysis based on detailed DSC models for vertical, oblique, and horizontal drops under accident scenarios, in accordance with International Atomic Energy Agency and U.S. Nuclear Regulatory Commission regulations. The results confirmed that ADCS significantly improved impact resistance and reduced plastic strain under impact conditions. Consequently, DSCs incorporating ADCS offer higher safety and reliability during both transportation and storage than those using conventional ASS. Furthermore, by leveraging these advantages of ADCS, DSCs can be made lighter while maintaining safety and structural integrity. Additionally, strain-based evaluations, consistent with ASME BPVC Sec. III Div. 3, demonstrated that higher margins could be achieved using ADCS where plastic deformation occurred. This highlights that ADCS is a promising material for ensuring greater structural integrity and safety in DSCs.
AToM: Adaptive Token Merging for Efficient Acceleration of Vision Transformer
No full textRecently, Vision Transformers (ViTs) have set anew standard in computer vision (CV), showing unparalleledimage processing performance. However, their substantial com-putational requirements hinder practical deployment, especiallyon resource-limited devices common in CV applications. Tokenmerging has emerged as a solution, condensing tokens withsimilar features to cut computational and memory demands.Yet, existing applications on ViTs often miss the mark in tokencompression, with rigid merging strategies and a lack of in-depth analysis of ViT merging characteristics. To overcome theseissues, this paper introduces Adaptive Token Merging (AToM), acomprehensive algorithm-architecture co-design for acceleratingViTs. The AToM algorithm employs an image-adaptive, fine-grained merging strategy, significantly boosting computationalefficiency. We also optimize the merging and unmerging processesto minimize overhead, employing techniques like First-Come-First-Merge mapping and Linear Distance Calculation. On thehardware side, the AToM architecture is tailor-made to exploit theAToM algorithm's benefits, with specialized engines for efficientmerge and unmerge operations. Our pipeline architecture ensuresend-to-end ViT processing, minimizing latency and memoryoverhead from the AToM algorithm. Across various hardwareplatforms including CPU, EdgeGPU, and GPU, AToM achievesaverage end-to-end speedups of 10.9x, 7.7x, and 5.4x, alongsideenergy savings of 24.9x, 1.8x, and 16.7x. Moreover, AToMoffers 1.2x1.9xhigher effective throughput compared toexisting transformer accelerators.
DFT-CES2: Quantum Mechanics Based Embedding for Mean-Field QM/MM of Solid-Liquid Interfaces
No full textThe solid-liquid interface plays a crucial role in governing complex chemical phenomena, such as heterogeneous catalysis and (photo)electrochemical processes. Despite its importance, acquiring atom-scale information about these buried interfaces remains highly challenging, which has led to an increasing demand for reliable atomic simulations of solid-liquid interfaces. Here, we introduce an innovative first-principles-based multiscale simulation approach called DFT-CES2, a mean-field QM/MM method. To accurately model interactions at the interface, we developed a quantum-mechanics-based embedding scheme that partitions complex noncovalent interactions into Pauli repulsion, Coulomb (including polarization), and London dispersion energies, which are described using atom-dependent transferable parameters. As validated by comparison with high-level quantum mechanical energies, DFT-CES2 demonstrates chemical accuracy in describing interfacial interactions. DFT-CES2 enables the investigation of complex solid-liquid interfaces while avoiding extensive parametrization. Therefore, we expect DFT-CES2 to be broadly applicable for elucidating atom-scale details of large scale solid-liquid interfaces for multicomponent systems.
LLDiffusion: Learning degradation representations in diffusion models for low-light image enhancement
No full textCurrent deep learning methods for low-light image enhancement typically rely on pixel-wise mappings using paired data, often overlooking the specific degradation factors inherent to low-light conditions, such as noise amplification, reduced contrast, and color distortion. This oversight can result in suboptimal performance. To address this limitation, we propose a degradation-aware learning framework that explicitly integrates degradation representations into the model design. We introduce LLDiffusion, a novel model composed of three key modules: a Degradation Generation Network (DGNET), a Dynamic Degradation-Aware Diffusion Module (DDDM), and a Latent Map Encoder (E). This approach enables joint learning of degradation representations, with the pre-trained Encoder (E) and DDDM effectively incorporating degradation and image priors into the diffusion process for improved enhancement. Extensive experiments on public benchmarks show that LLDiffusion outperforms state-of-the-art low-light image enhancement methods quantitatively and qualitatively. The source code and pre-trained models will be available at https://github.com/TaoWangzj/LLDiffusion.
Optimized UAV view planning for high-quality 3D reconstruction of buildings using a modified sparrow search algorithm
No full textHigh-quality 3D reconstruction of existing buildings is essential for their maintenance, restoration, and management. Effective view planning for image collection significantly impacts the quality of photogrammetry-based 3D reconstruction. Intricate building structures, such as the overhangs, protrusions, and concave regions, can lead to under-sampled regions with traditional view planning methods, while excessively increasing the number of views require substantial computational resources and data collection efforts. To address these issues, this paper proposes a novel exploration-then-exploitation view planning strategy to achieve high-quality building reconstruction with minimal views. Firstly, the UAV no-fly regions and building attention regions are identified through semantic and geometric analysis of the images and coarse model during the exploration stage. Then, a novel optimization fitness function is mathematically formulated, considering building attention regions and reconstruction influential factors, including distance, incidence angle, parallax angle, and overlap. Furthermore, a modified sparrow search algorithm is proposed with the improved optimization mechanism and the integration of view planning physical model, enabling effective generation of optimal viewpoint set. Finally, the collisionfree shortest trajectory is designed, allowing the UAV to collect images and reconstruct a high-quality model during exploitation stage. Experiments in virtual and real-world scenarios validate the effectiveness of our proposed modified SSA mechanism and the view planning strategy. Results demonstrate that the modified SSA achieves higher convergence accuracy and speed compared to the original SSA, PSO and GA. Our strategy can generate more accurate and complete 3D reconstruction models with the same or fewer captured images compared to commonly used and state-of-the-art strategies.
SOUND ABSORBING STRUCTURE THAT CAN HARVEST ENERGY AND METHOD OF MANUFACTURING THE SAME
No full text본 발명은 에너지 수확이 가능한 흡음구조체 및 이를 제조하는 방법에 관한 것으로서, 압전고분자를 용매에 용해하는 압전 고분자 용해단계; 제1발포성 고분자에 용매를 첨가하는 용매 첨가단계; 압전 고분자 용해단계에서 생성된 제1용액과 용매 첨가단계에서 생성된 제2용액을 혼합하여 제1혼합물을 생성하는 제1혼합단계; 제1혼합물에 제2발포성 고분자를 혼합하여 제2혼합물을 생성하는 제2혼합단계; 제2혼합물에 마이크로파를 조사하여 발포가 수행되도록 하는 마이크로파 조사단계; 및 고화(固化)된 압전고분자로부터 잔류한 용매를 제거하여 미세기공을 형성시키는 미세기공 형성단계;를 포함하는, 에너지 수확이 가능한 흡음구조체를 제조하는 방법이 제공된다
NEW POLYMER DONOR, COMPOSTION FOR ORGANIC ELECTRONIC DEVICE AND ORGANIC ELECTRONIC DEVICE
No full text본 발명은, 신규한 고분자 주개 화합물, 이를 포함하는 유기전자소자용 조성물 및 유기전자소자에 관한 것으로, 보다 구체적으로, 화학식 1로 표시되는 화합물인 신규한 고분자 주개 화합물, 이를 포함하는 유기전자소자용 조성물 및 이를 포함하는 유기태양전지에 관한 것이다
MULTICAST SWITCH ARRAY AND BEAM STEERING APPARATUS USING SAME
No full text멀티캐스트 스위치 어레이(Multicast Switch Array) 및 이를 이용한 멀티빔(multibeam) 조향 장치가 제공된다. 이로써 대규모 멀티빔을 정밀하게 조향가능하다
METHOD FOR PREPARING CATALYST FOR ELECTROCHEMICAL CARBON DIOXIDE REDUCTION
No full text구리(Cu)가 유기 리간드로 연결되고 코발트(Co)가 도핑된 금속유기골격체(MOF)를 이용하여, 전기화학적 이산화탄소 환원 반응의 선택성과 수율을 모두 높일 수 있는 촉매를 제조할 수 있다. 또한 상기 촉매에서 구리 중의 코발트의 도핑량을 다양하게 조절함으로써 전기화학적 이산화탄소 환원 특성을 필요에 맞게 제어할 수 있다
Method And Apparatus for GNN-Acceleration for Efficient Parallel Processing of Massive Datasets
No full text대규모 그래프 데이터의 효율적인 병렬 처리를 위한 그래프 신경망 가속 방법 및 장치를 개시한다. 본 개시의 일 측면에 의하면, 대규모 그래프 데이터의 효율적인 병렬 처리를 위한 그래프 신경망 가속 장치로서, 레이어별 서브그래프 및 임베딩 테이블을 획득하여, 특징 차원 및 스트리밍 멀티프로세스(이하 'SM') 내 최대 쓰레드 수에 기초하여 하나의 목적지-버텍스의 임베딩 처리를 위해 할당할 상기 SM의 수를 결정하고, 상기 서브그래프에 포함된 모든 목적지-버텍스들 각각에 상기 결정된 수의 SM을 할당하는 SM스케줄러; 및 상기 SM이 자신에게 할당된 목적지-버텍스의 임베딩을 획득하고, 상기 SM이 상기 서브그래프를 이용하여 상기 목적지-버텍스의 적어도 하나 이상의 이웃-버텍스의 임베딩을 획득하여, 상기 SM이 상기 목적지-버텍스 및 상기 이웃-버텍스의 임베딩들을 이용하여 사용자 지정 연산을 수행하는 연산부를 포함하는 그래프 신경망 가속을 위한 장치 및 그의 동작 방법을 제공한다