A5023573950- High voltage insulation and dielectric phenomena
- Opportunistic and Delay-Tolerant Networks
- Power Transformer Diagnostics and Insulation
- Satellite Communication Systems
- Advanced Neural Network Applications
- Video Surveillance and Tracking Methods
- Advanced Algorithms and Applications
- Dental Radiography and Imaging
- Topic Modeling
- Fire Detection and Safety Systems
- Age of Information Optimization
- Natural Language Processing Techniques
- Electrostatic Discharge in Electronics
- Advanced Image and Video Retrieval Techniques
- Advanced Fiber Optic Sensors
- Underwater Acoustics Research
- Scheduling and Optimization Algorithms
- Lightning and Electromagnetic Phenomena
- 3D Shape Modeling and Analysis
- Advanced Manufacturing and Logistics Optimization
- Orthodontics and Dentofacial Orthopedics
- Domain Adaptation and Few-Shot Learning
- Image and Object Detection Techniques
- Medical Imaging and Analysis
- Optimization and Packing Problems
Academy of Military Medical Sciences
2025
Shanghai Electric (China)
2025
Second Hospital of Shanxi Medical University
2024
Shanxi Medical University
2024
Zhuhai Institute of Advanced Technology
2024
Beijing Institute of Technology
2024
Xi'an University of Technology
2024
Kunming University of Science and Technology
2024
Anhui Medical University
2024
China Electronics Technology Group Corporation
2022-2024
The classification and recognition technology of underwater acoustic signal were always an important research content in the field processing. Currently, wavelet transform, Hilbert-Huang Mel frequency cepstral coefficients are used as a method feature extraction. In this paper, for extraction identification noise data based on CNN ELM is proposed. An automatic signals proposed using depth convolution network. target classifier extreme learning machine. Although neural networks can execute...
Unsupervised domain adaptation (UDA) assumes that source and target data are freely available usually trained together to reduce the gap. However, considering privacy inefficiency of transmission, it is impractical in real scenarios. Hence, draws our eyes optimize network without accessing labeled data. To explore this direction object detection, for first time, we propose a data-free adaptive detection (SFOD) framework via modeling into problem learning with noisy labels. Generally,...
Copy module has been widely equipped in the recent abstractive summarization models, which facilitates decoder to extract words from source into summary. Generally, encoder-decoder attention is served as copy distribution, while how guarantee that important are copied remains a challenge. In this work, we propose Transformer-based model enhance mechanism. Specifically, identify importance of each word based on degree centrality with directed graph built by self-attention layer Transformer....
In this paper, a digital dc PD pulse detection system with bandwidth of 10 kHz - 40 MHz is introduced, which was developed using some artificial intelligence methodologies. Focus made on detection, grouping and classification random signals generated by phenomena at voltage. Digital only resorting to band-pass filter, high-speed digitizer (100 MS/s) PC data processing software. Grouping realized feature extraction waveshapes equivalent time-frequency method (ETFM), making the 2D parameters...
We present an abstractive summarization system that produces summary for Chinese e-commerce products. This task is more challenging than general text summarization. First, the appearance of a product typically plays significant role in customers' decisions to buy or not, which requires model effectively use visual information product. Furthermore, different products have remarkable features various aspects, such as “energy efficiency” and “large capacity” refrigerators. Meanwhile, customers...
In the field of defect detection, image processing algorithms and feature extraction have some limitations, owing to their necessity for extracting a large number different features diverse products images. Meanwhile, images defective are less various. Aiming at these problems, we presented One-Class classifier based on deep convolution neural network detect in this paper. We design loss function with penalty term Euclidean distance train model. A hypersphere is used as classification...
Gait phase detection is a new biometric method which of great significance in gait correction, disease diagnosis, and exoskeleton assisted robots. Especially for the development bone robots, recognition an indispensable key technology. In this study, main characteristics phases were determined to identify each phase. A long short-term memory-deep neural network (LSTM-DNN) algorithm proposed gate detection. Compared with traditional threshold LSTM, has higher accuracy different walking speeds...
According to the advantages and disadvantages of electromagnetic acoustic partial discharges (PD) measurements shown in IEC TS 62478, a novel integrated sensor with acoustical emission (AE) ultrahigh frequency (UHF) methods for PD detection transformers is developed this article. It formed by placing an AE inside end part UHF probe, can be mounted on oil valve, pushed transformer measure signals simultaneously. The sensitivity validity are proved experimental tests physical 110-kV...
Cephalometric analysis relies on accurate detection of craniomaxillofacial (CMF) landmarks from cone-beam computed tomography (CBCT) images. However, due to the complexity CMF bony structures, it is difficult localize efficiently and accurately. In this paper, we propose a deep learning framework tackle challenge by jointly digitalizing 105 CBCT By explicitly local geometrical relationships between landmarks, our approach extends Mask R-CNN for end-to-end prediction landmark locations....
Abstract To improve heat dissipation performance of panel-type radiator for transformers, this study investigated the flow and transfer characteristics in air-side metal foam partially filled channels radiator. The porous thin-layer (PTLF) fin (PFF) methods filling ratio ( V p ) were analyzed compared. result indicated that permeability interfacial turbulent kinetic energy region PFF channel are higher. Increasing can promote mixing transfer. For Re = 5,125–15,375, when 11.1%, evaluation...
To improve the thermal performance of air-cooled panel-type radiators for transformers, a multi-fan horizontal blowing method was designed in this paper, and thermo-hydraulic oil-side air-side radiator investigated with simplified numerical experiments. The uniform air distribution zoned heat dissipation ideas were used three methods, which can increase proportion supply high-temperature area apply multiple fans insulating oil radiator. Then, effect different flow rates on investigated. It...
In this paper, we consider the problem of routing in disruption-tolerant-networking-based earth-observing satellite networks, which are characterized by a frequently changing topology and potentially sparse intermittent connectivity. To handle challenges posed these properties, propose joined space-temporal algorithmic framework for those where time-varying is modeled as space-time graph leveraging predictability satellites' relative motions. Based on model, devise multipath algorithm...
Integrated Satellite-Terrestrial (IST) networks consisting of low earth orbit (LEO) satellite constellation and terrestrial users are widely developed for potentially attractive requirements content distribution, benefiting from deployment cost broader coverage capability. However, orbital paradigm LEO constellation, as well an accordingly time-changing topology, makes a huge challenge distributing objective file due to heavy lack end-to-end paths. To address this issue, in paper,...
Extrinsic Fabry⁻Perot (FP) interferometric sensors are being intensively applied for partial discharge (PD) detection and localization. Previous research work has mainly focused on novel structures materials to improve the sensitivity linear response of these sensors. However, directional behavior an FP ultrasonic sensor is also particular importance in localizing PD source, which rarely considered. Here, a microelectromechanical system (MEMS)-based with 5-μm-thick micromechanical...
It is clinically important to accurately predict facial soft-tissue changes prior orthognathic surgery. However, the current simulation methods are problematic, especially in anatomic regions of clinical significance, e.g., nose, lips, and chin. We developed a new 3-stage finite element method (FEM) approach that incorporates realistic tissue sliding improve such prediction.In Stage One, change was simulated, using FEM with patient-specific mesh models generated from our previously eFace...
In this paper, a video fire detection method is proposed, which demonstrated good performance in indoor environment. Three main novel ideas have been introduced. Firstly, flame color model RGB and HIS space used to extract pre-detected regions instead of traditional motion differential method, as it more suitable for Secondly, according the flicker characteristic flame, similarity two values centroid are proposed. At same time, simple but effective tracking consecutive frames established....
This study describes a novel fiber optic extrinsic Fabry–Perot interferometric (EFPI) ultrasonic sensor comprising low-cost and high-performance silicon diaphragm. A vibrating diaphragm, 5 μm thick, was fabricated by using the Microelectromechanical Systems (MEMS) processing technology on silicon-on-insulator (SOI) wafer. The (FP) cavity length solely determined during manufacturing process of diaphragm defining specific stepped hole handling layer SOI wafer, which made assembly easier. In...