A5101676902- Advanced Neural Network Applications
- Distributed and Parallel Computing Systems
- Image Processing Techniques and Applications
- IoT and Edge/Fog Computing
- Software-Defined Networks and 5G
- Satellite Communication Systems
- Advanced Image and Video Retrieval Techniques
- Digital Image Processing Techniques
- Bluetooth and Wireless Communication Technologies
- Advanced Image Processing Techniques
- Domain Adaptation and Few-Shot Learning
- Vacuum and Plasma Arcs
- Parallel Computing and Optimization Techniques
- Wireless Body Area Networks
- Model Reduction and Neural Networks
- GNSS positioning and interference
- Nuclear Engineering Thermal-Hydraulics
- Complex Network Analysis Techniques
- Particle accelerators and beam dynamics
- Advanced Graph Neural Networks
- Educational Technology and Pedagogy
- Network Time Synchronization Technologies
- Anomaly Detection Techniques and Applications
- Electrical Contact Performance and Analysis
- Caching and Content Delivery
Jilin Electric Power Research Institute (China)
2025
Anhui Agricultural University
2023-2024
Beijing Microelectronics Technology Institute
2020-2024
Nanjing University of Aeronautics and Astronautics
2024
Beijing Institute of Technology
2019-2023
Harbin Institute of Technology
2022
Ningbo City College of Vocational Technology
2008-2020
Zhejiang Normal University
2020
Shanghai Jiao Tong University
2010
Objective: To construct a classified evaluation model of middle school students’ physical health based on biomechanical parameters and to achieve accurate intervention adolescents’ through biomechanically informed strategies. Methods: Cluster sampling was used select 1261 male samples from eight junior schools in Jinhua City, Zhejiang Province 2022 2023 as data set. Second-order clustering BP neural network algorithm were establish classification for students, the accuracy verified. Results:...
The exponential growth in user numbers has resulted an overwhelming surge data that the smart grid must process. To tackle this challenge, edge computing emerges as a vital solution. However, current heuristic resource scheduling approaches often suffer from fragmentation and consequently get stuck local optimum solutions. This paper introduces novel network allocation method for multi-domain virtual networks with support of computing. approach entails modeling model formulating constraints...
As deep learning moves toward edge computing, researchers have developed techniques for efficient resource usage and accurate inference on mobile devices. Quantization, as one of the key approaches, enables deployment models embedded platforms. However, MobileNet's accuracy suffers due to quantization errors in depth-wise separable convolutions. To reach a smaller model size, we turn mixed-precision strategy instead uniform quantization. Motivated gain higher precision, quantization-friendly...
Origin-destination (OD) crowd flow, if more accurately inferred at a fine-grained level, has the potential to enhance efficacy of various urban applications. While in practice for mining OD flow with effect, problem spatially interpolating occurs since ineluctable missing values. This is further complicated by inherent scarcity and noise nature data. In this paper, we propose an uncertainty-aware interpolative explainable framework, namely UApex, realizing reliable trustworthy interpolation....
With the massive access to sensing, measuring, and user-side intelligent terminal devices, functionalities applications at edge-side of digital distribution networks have become significantly enriched. However, limited computation resources edge computing devices must be utilized efficiently achieve various functions, presenting challenges resource scheduling within networks. To tackle these challenges, this paper proposes an optimal method for multi-functional devices. The collaborative...
Xilinx and Mathworks jointly proposed System Generator (SysGen) Simulink to accelerate development of DSP (digital signal processing) style applications on Field Programmable Gate Array (FPGA) chips. However, most developments with SysGen end at simulation stage without complete stand-alone implementation FPGA since these tools do not come up sufficient IO system libraries. In this paper, we present the process implementing a full reconstruct CT (computed tomography) images using SysGen....
FPGAs have been improved significantly in terms of performance and capacity over the last 20 years. The scale FPGA based design also sparked off demands for high-level synthesis to handle complicated applications. A well known intricate application is FMM (Fast Multipole Method) algorithm N-body problem, which so that it was not implemented on as reported literature. In this paper, we take high level modeling tools, i.e. Simulink System Generator implement major modules solve N-Body problem...
This paper designs a hardware acceleration system based on FPGA and deep learning algorithm for image recognition tasks. While implementing the convolutional neural network ZYNQ series chip, real-time architecture is designed. The ARM embedded in used to collect, store display input data real time. collected stored transmitted through AXI4 bus hardwareized complete of image. experimental results show that can 528 convolution operations single clock cycle. After 11-bit fixed-point...
This paper proposes a data anomaly detection and correction algorithm for the tea plantation IoT system based on deep learning, aiming at multi-cause multi-feature characteristics of abnormal data. The is Z-score standardization original determination sliding window size according to sampling frequency. First, we construct convolutional neural network (CNN) model extract Second, support vector machine (SVM) algorithm, Gaussian radial basis function (RBF) one-to-one (OVO) multiclassification...
Bridge structures, together with magnetic blowing components are frequently used in power relays or contactors for switching heavy DC load. However, the asynchronous mechanical separation of double multiple contact pairs series makes arc interruption behavior more complex. In order to improve capability electromagnetic relay, effects force and field strength on electrical lifetime investigated separately. During endurance experiments, current voltage waveforms recorded explicitly, featured...
With the rapid development of wireless sensor networks, precision time synchronization between nodes is becoming increasingly attractive. However, low cost, power, technology and other requirements restrict use algorithms in networks. TPSN algorithm a two-way algorithm, which widely used network. But due to media access mechanism channel stability network, standard can not meet requirement high synchronization. In this paper, have been improved based on doze mechanism. Under control central...
Personal area network has been widely used in embedded application areas. The paper indicates structure and implementation of wireless storage equipment, pocket server, which acts as a data center the PAN. Bluetooth technology is to support communication. Beyond basic implementation, cost performance issues well valued functions are concerned.
Ionosphere is an important part of the satellite-ground radio links, and dispersive characteristic most factor affecting wave propagation in background ionosphere. A method presented to simulate dynamic ionospheric characteristic. In method, according simulated scenario, total electron content (TEC) links calculated by NeQuick2 ionosphere density model. Based on nonlinear phase time-varying filtering system, distortion wideband signal caused dispersion realized. The simulation results show...
In order to solve the information transfer of social network media, a new prediction method is presented by triangle ring attractor. At first, description media given in this model, and evaluation indexes such as recall rate, matching degree, rate are emphatically expounded. same time, characteristics analyzed with triangular attractor, an model established. Finally, through simulation experiment, key factors influencing deeply analyzed. Experimental results show that compared other...
Abstract In order to enhance the visual effects of reconstructed images in single image super-resolution, and solve problem instability training phase super-resolution with generative adversary network (SRGAN), a new model is built more depth width, corresponding algorithm proposed. The structure original SRGAN modified, so that deeper wider convolutional networks can be used high efficiency. Furthermore, loss function employed balance discriminator network, phase. for are obtained from...