A5088699621- Machine Fault Diagnosis Techniques
- Network Time Synchronization Technologies
- Energy Efficient Wireless Sensor Networks
- Distributed Control Multi-Agent Systems
- Nonlinear Dynamics and Pattern Formation
- Indoor and Outdoor Localization Technologies
- Fault Detection and Control Systems
- Sparse and Compressive Sensing Techniques
- Neural Networks Stability and Synchronization
- Anomaly Detection Techniques and Applications
- Gear and Bearing Dynamics Analysis
- Engineering Diagnostics and Reliability
- Smart Grid and Power Systems
- Optical measurement and interference techniques
- Image Processing Techniques and Applications
- Image and Signal Denoising Methods
- Advanced Memory and Neural Computing
- Advanced Measurement and Detection Methods
- Modular Robots and Swarm Intelligence
- Surface Roughness and Optical Measurements
- Generative Adversarial Networks and Image Synthesis
- IoT and Edge/Fog Computing
- Power Systems Fault Detection
- Advanced Neural Network Applications
- Microwave Imaging and Scattering Analysis
Beijing Institute of Technology
2014-2022
China Aerospace Science and Industry Corporation (China)
2022
Fire Safety Design
2022
Institute of Navigation
2022
Ministry of Education of the People's Republic of China
2017
Prognostics and health management allows us to predict the remaining useful life (RUL) of machinery, which is important in reducing maintenance costs downtime, even preventing casualties. Bearing faults account for a large proportion machine faults. To RUL bearings, indicators that represent degeneration state are extracted based on Hilbert-Huang transform selected according Spearman's coefficient. A model-based particle filter method then used track degradation state. The unknown parameters...
With the widespread application of wireless sensor networks, large-scale systems with high sampling rates are becoming more and common. The amount original data generated by network is very large, transmitting all back to host wastes bandwidth energy. This paper proposes a transmission method for large based on hierarchical compressed sensing sparse decomposition. includes signal decomposition same basis different mask. Compared traditional method, this reduces error reconstruction,...
Time synchronization is a crucial component of WSN. As the rapid development mesh network, firefly-inspired algorithm comes out for solving issue large-scale networks. But it usually has short cycle improving sync rate. In this paper, multiscale strategy proposed based on long cycle. First, firefly model used in our introduced. Then description about synchronization. Simulation results show great acceleration rate and improvement stability.
In recent years, the use of wireless sensor networks has become increasingly widespread. Because instability networks, packet loss occasionally occurs. To reduce impact on data integrity, we take advantage deep neural network's excellent ability to understand natural and propose a repair method based convolutional network with an encoder-decoder architecture. Compared common interpolation algorithms compressed sensing algorithms, this obtains better results, is suitable for wider range...
Data acquisition is the foundation of soft sensor and data fusion. Distributed its synchronization are important technologies to ensure accuracy sensors. As a research topic in bionic science, firefly-inspired algorithm has attracted widespread attention as new method. Aiming at reducing design difficulty algorithms for Wireless Sensor Networks (WSNs) with complex topologies, this paper presents based on multiscale discrete phase model that can optimize performance tradeoff between network...
Abstract Integrated drive-line fluids play a role in transmitting torque, providing control pressure, lubrication, and cleaning the integrated drive-train. The increase of oil abrasive particles will accelerate wear rate rotating parts, while insufficient lead to rapid ablation gluing parts. According statistics, more than 75% hydraulic equipment failure is caused by pollution. Oil pollution leads filter blockage which cause fatal so that entire transmission system cannot work. Therefore, it...
Abstract In order to improve the generalization ability of model under different working conditions and robustness intelligent fault diagnosis, learn a broader feature representation, this paper proposes an diagnosis method that integrates condition attribute encoding multi-scale cascade concepts. This information into vibration data by introducing methods such as coding, modules quadruple losses, effectively extracts invariant features. trains classification through three-stage training...
Synchronicity is the base of sleep and superframe, used by many protocols in WSN. Distributed synchronicity algorithm become popular rubust WSN, due to unstratified topological structure low computational complexity. The firefly-inspired (FSA), as typical distributed synchronicity, usually needs a large coupling coefficient for accelerating convergence. However, may lead instability inaccuracy. So, this paper proposes novel self-adaption algorithm, aiming at fast accurate synchronicity. This...
As the continuous development of wireless sensor networks, network has put forward higher requirements for transmission large amounts data.In order to meet high efficiency transmission, we often use various technologies increase rate, but this may also cause packet loss rate on network.Therefore, in reduce network, paper proposes a lost retransmission technology based reliable data networks.This is multi-hop efficient platform composed multiple nodes.This an FPGA-based mechanism that can...
Aiming at the problem that remaining useful life (RUL) prediction algorithm of a rolling bearing produces large amount power consumption in long-term operation edge devices, RUL method using variable interval sampling is proposed. Firstly, extract time domain features, frequency features and abstract convolutional neural networks (CNNs) original vibration signal bearing. After normalization, are selected according to certain criteria, then input into long short-term memory (LSTM) network...
With the development of information age, importance edge computing has been highlighted in industrial site monitoring, health management, and fault diagnosis. Among them, processing signals scenarios is cornerstone realizing these scenarios. While performance devices dramatically improved, demand for signal side also ushered explosive growth. However, deployment traditional serial or parallel architectures on problems such as poor flexibility, low efficiency, resource utilization, making...
In test-range tracking, traditional machine vision systems are composed of a frame grabber and host PC, which cannot meet the requirement real-time processing. this paper, we present locating system based on FPGA (Field Programmable Gate Array) to solve problems caused by high-speed low-sharpness target in field testing. FPGA, as processing core, captures high resolution images from TDI-CCD (Time delay integration charge coupled device) line scan camera is Camera Link interface. Two SRAM...
This paper proposes a telemetry data transmission method based on wireless multi-hop network.This is network constituted by several nodes and adopts modified ZigBee communication protocol.In this method, the terminal gateway will coordinate whole automatically during order transmission.According to experiment results, can guarantee stable high-efficient be used in many scenario.
With the rapid development of wireless sensor network, networks become more complex. Original synchronicity algorithms may not work efficiently for complex mesh network. In this paper, we present a new firefly-inspired stochastic coupled strategy with simplified discrete phase model in It is software-based strategy, which can fully utilize communication bandwidth and be insensitive to packet loss rate. First, paper presents details. The stability evaluated effects varying parameters are...
Aiming at the problem of difficulty in detection small targets large field view, an image processing algorithm is proposed to detect view.The first performs preprocessing using grayscale transformation and median filtering, then Gabor matched filtering threshold segmentation are used obtain binary image.For getting position target image, centroid presented.Finally, validated by practical experiments.The results show that can effectively targets.
A vibration data repair method based on Generative Adversarial Networks (GAN) is proposed to resolve the problem of incomplete acquisition bearing under certain circumstances (sensor failure, extreme environments, etc.), which leads errors in analysis. We use a GAN framework, combined with an Auto Encoder (AE), become Encoder-Generative (AE-GAN) generate synthetic related interpolation. First, introduced generator reconstruct input missing by encoding and decoding. Then, reconstructed...
With the increase of fault history data, problem high precision and long-time prediction under different failure modes is presented. We propose a multi-channel fusion algorithm based on Long Short-Term Memory (LSTM) deep network. The ability increases with training samples. Based analysis influence network parameters accuracy, optimal are selected to realize high-precision prediction. It can recognize without historical data. And it integrate information for off-line achieve goal...