A5102937553- Parallel Computing and Optimization Techniques
- Distributed and Parallel Computing Systems
- Industrial Technology and Control Systems
- Advanced Algorithms and Applications
- Low-power high-performance VLSI design
- Advanced Data Storage Technologies
- Power Systems and Renewable Energy
- Lattice Boltzmann Simulation Studies
- Fluid Dynamics and Heat Transfer
- Power Systems and Technologies
- Remote Sensing and Land Use
- Coal and Coke Industries Research
- Energy Efficient Wireless Sensor Networks
- Icing and De-icing Technologies
- Thermochemical Biomass Conversion Processes
- Safety and Risk Management
- Vehicular Ad Hoc Networks (VANETs)
- EEG and Brain-Computer Interfaces
- Coal Combustion and Slurry Processing
- Remote-Sensing Image Classification
- Non-Invasive Vital Sign Monitoring
- Advanced Computing and Algorithms
- Fluid Dynamics Simulations and Interactions
- Integrated Energy Systems Optimization
- Advanced Image Fusion Techniques
Zhengzhou University
2020-2024
Nanyang Institute of Technology
2018-2023
Zhengzhou University of Aeronautics
2023
Tongji University
2023
Chinese Academy of Sciences
2023
Research Institute for Electromagnetic Materials
2014
Hebei University of Engineering
2009
Electric Power University
2005
CAE (Canada)
2001
The application of deep neural networks (DNNs) for road extraction from remote sensing images has gained broad interest because the competence concerning complex nonlinear relations; however, presence noisy labels in training data sets adversely affects performance DNNs. existing methods improving robustness DNNs focus on modeling noise distribution. However, these approaches are not satisfactory inaccurate high-level image features obtained by To address this issue, we develop a...
Abstract Energy distribution networks represent crucial infrastructures for modern society, and various simulation tools have been widely used by energy suppliers to manage these intricate networks. However, calculations include a large number of fluid control equations, computational overhead limits the performance software. This paper proposes universal parallel framework pipeline that takes advantages data parallelism independence between network elements. A non-pipe model an supply is...
To solve the problem of fault diagnosis for engine, due to complexity equipments and particularity operating environments, generally speaking, there is no one-to-one correspondence between characteristic parameters status, so, methods are very complicated. A novel method based on empirical mode decomposition (EMD) wavelet packet BP neural network proposed in this paper. Firstly, given signal analyzed by remove noise; Then de-noised data decomposed into a number IMFs EMD extract their...
The localization of continuous objects and the scheduling resources are challenging issues in wireless sensor networks (WSNs). Due to irregular shape target area deployment WSNs, data always discrete sparse, most network also limited by node energy. To achieve faster detection tracking objects, we propose a convolution-based object algorithm (named CCOL). Moreover, implement idea greedy dynamic programming design an energy-saving efficient strategy model MSSM) respond emergencies caused...
The Internet of Things (IoT) faces significant challenges in the requirements sensitive task latency, reasonable resource allocation and reliability for transactions. This paper introduces a novel method road IoT context connected autonomous electric vehicles (CAEVs). proposed algorithm leverages ant colony (ACA) to effectively allocate coordinate resources within groups CAEVs. By considering energy consumption pheromone volatilization, coordination process are optimized. To improve linear...
DCU (Deep Computing Unit) acceleration device is a kind of GPU-like device, which developed by GPU technology authorized AMD, so it suitable to solve the parallel problems many data elements in same program. In this paper, immersed boundary method based on was studied realize efficient two-phase flow simulation. The Jacobi multigrid used pressure Poisson equation simulations. Based our serial code and combined with hardware characteristics HIP (Heterogeneous Interface for Portability)...
The network of energy pipelines is extremely important for the proper functioning a modern society. Numerical simulation fluid networks offers rather easy and effective technique to track work performance pipelines. Because computational intensiveness pipeline network, this kind parallel may undoubtedly be enhanced if reasonably divided into several subnetworks in which flows are resolved independently. In paper, we concentrate on how achieve computationally balanced subnetwork partition...
As an emerging computing model, edge greatly expands the collaboration capabilities of servers. It makes full use available resources around users to quickly complete task request coming from terminal devices. Task offloading is a common solution for improving efficiency execution on networks. However, peculiarities networks, especially random access mobile devices, brings unpredictable challenges in network. In this paper, we propose trajectory prediction model moving targets networks...
<title>Abstract</title> Energy distribution networks represent crucial infrastructures for modern society, and various simulation tools have been widely used by energy suppliers to manage these intricate networks. However, calculations include a large number of fluid control equations, computational overhead limits the performance software. This paper proposes universal parallel framework pipeline that takes advantages data parallelism independence between network elements. A non-pipe model...
Han, P. and Y., 2018. Noise reduction recovery algorithm for underwater distorted image. In: Liu, Z.L. Mi, C. (eds.), Advances in Sustainable Port Ocean Engineering. Journal of Coastal Research, Special Issue No. 83, pp. 720–723. Coconut Creek (Florida), ISSN 0749-0208.Due to some unavoidable internal factors digital cameras other imaging equipment, as well human environmental external factors, the photographing images will be distorted, thus affecting recognition effect optical character...
SpMV(Sparse matrix-vector multiplication) plays an important role in the fields of scientific research and engineering, but its non-merged memory access load imbalance cause worse performance when dealing with very large-scale sparse matrices. In view above problems, this paper proposes a SpMV algorithm MPI+HIP hybrid model based on DCU accelerator, which uses on-chip high-speed storage space at fine-grained level to reduce hidden latency global memory. Through warp reduction, barriers...
In different flow fields, the description and tracking of droplets in computer has high time complexity. order to improve efficiency droplet simulation calculation, on domestic DCU acceleration device, immersed boundary methods is used realize deformation under shear flow. After finite difference hot spot pressure Poisson equation, Jacobian implicit iterative method for parallel improvement, divide a larger computing task into multiple independent modules, make reasonable use local data...
In order to give full play the advantages of DCU accelerator and solve problems algorithm SpMV (Sparse matrix-vector multiplication) with limited bandwidth, unbalanced load, non-combined memory access, a SCSR (Static Compressed Sparse Row) using CSR (Compressed storage format is proposed based on accelerator. The statically allocates same number rows each thread block according average non-zero elements in row avoid unnecessary computations; application resources reduced by reusing on-chip...