A5056787653- Advanced Neural Network Applications
- Energy Load and Power Forecasting
- Industrial Vision Systems and Defect Detection
- Advanced Machining and Optimization Techniques
- Advanced machining processes and optimization
- Tribology and Wear Analysis
- Remote Sensing and LiDAR Applications
- Advancements in Photolithography Techniques
- Solar Radiation and Photovoltaics
- Image and Object Detection Techniques
- Elevator Systems and Control
- Infrastructure Maintenance and Monitoring
- Stock Market Forecasting Methods
- Advanced Image Processing Techniques
- Smart Grid Energy Management
- Power Line Inspection Robots
- Video Surveillance and Tracking Methods
- Advanced Surface Polishing Techniques
- Generative Adversarial Networks and Image Synthesis
- Image and Signal Denoising Methods
- Viral Infectious Diseases and Gene Expression in Insects
- Smart Grid and Power Systems
- Currency Recognition and Detection
- Structural Health Monitoring Techniques
- Microgrid Control and Optimization
Chongqing University
2022-2024
South China University of Technology
2024
Xi'an Jiaotong University
2022-2024
McMaster University
2024
State Key Laboratory of Electrical Insulation and Power Equipment
2023
Non-intrusive load monitoring constitutes a significant function of the smart grid in future. The purpose is to ameliorate consumption and supply electricity by disaggregating total appliance-level without intrusive monitoring. Recently, energy disaggregation improved with emergence deep learning, but imbalanced datasets long sequences bring multiple difficulties model training. distribution on/off states limitation lead massive false positive samples undetected events. To tackle these...
Accurate load forecasting can maintain the safety and stability of power grids. The mainstream models are based on complex recurrent or convolutional neural networks (RNNs, CNNs), in recent years they often used combination with attention mechanism. shortcomings these that cannot get rid sequential computation fail to capture long-term dependence. For better application day-ahead (DALF), we propose a new network architecture, i.e., Forwardformer, which is implemented by imposing some...
Abstract Machine learning (ML) constructs predictive models by understanding the relationship between protein sequences and their functions, enabling efficient identification of with high fitness values without falling into local optima, like directional evolution. However, how to extract most pertinent functional feature information from a limited number is vital for optimizing performance ML models. Here, we propose scut_ProFP (Protein Fitness Predictor), framework that integrates...
Abstract The particular work environment of transmission lines makes it liable to malfunction because the foreign bodies attached power equipment. However, traditional detection methods are difficult identify exact category and ignore relationship between a body equipment, which is correlated determining whether there fault or not. In order detect distinguish give warnings automatically in line patrol with high accuracy efficiency, we improve original YOLOv5 object model propose novel fusion...
In the realm of electrical engineering, object detection in infrared images is paramount for maintenance power equipment. However, scarcity image datasets severely impedes improvement technology images. To address this issue, we propose a data enhancement method based on Cycle-consistent Generative Adversarial Networks (CycleGAN), which can translate easily available visible equipment into less accessible and achieve to expand pool Because cycle-consistent loss function, CycleGAN realize...
Multi-turbine wind power (WP) prediction contributes to turbine (WT) management and refined farm operations. However, the intricate dynamic nature of interrelationships among WTs hinders full exploration their potential in improving prediction. This paper proposes a novel spatio-positional series attention long short-term memory (SPSA-LSTM) method, which extracts hidden correlations temporal features from speed (WS) WP historical data different for high-precision Using embedding techniques,...
In the rapidly evolving electronics manufacturing sector, maintaining quality control and conducting failure analysis of Printed Circuit Boards (PCBs) are critical yet challenging tasks. This study presents a groundbreaking self-supervised learning framework to address existing gaps in reconstruction encoded or blurred Board images. By leveraging customized DeepLabV3+ architecture with depth-wise separable convolutions, our model is engineered autonomously learn intrinsic features,...
Object detection techniques have experienced rapid development. However, challenges such as low recognition accuracy, high false-negative rates and slow speed still exist in the task of infrared for power equipment. To address these issues, this paper proposes a equipment image model (YOLO4IIR) based on attention mechanisms convolutional neural networks. By improving upon YOLOv5 architecture introducing EMO framework, CoTAttention, CARAFE operator, model's capability is enhanced while...
In order to reduce the risk power system from bird nests in distribution line towers, grids often use modern monitoring devices for real-time detection of bird's nests. However, traditional object models are typically difficult implement on mobile side and frequently involve big arithmetic cases. And some lightweight have poor reliability accuracy. Therefore, this paper proposes a Shuffle-BiFPN-YOLOv5 model, which lightens original YOLOv5 model fuses more features perform accurate quick nest...