A5100658188- Machine Fault Diagnosis Techniques
- Advanced Algorithms and Applications
- Advanced Sensor and Control Systems
- Image and Signal Denoising Methods
- COVID-19 diagnosis using AI
- Speech and Audio Processing
- Engineering Diagnostics and Reliability
- Domain Adaptation and Few-Shot Learning
- Indoor and Outdoor Localization Technologies
- Quantum Information and Cryptography
- Quantum Mechanics and Applications
- Advanced Measurement and Detection Methods
- Anomaly Detection Techniques and Applications
- Neural Networks and Applications
- Fault Detection and Control Systems
- Advanced Computing and Algorithms
- Neural Networks and Reservoir Computing
- Advanced Combustion Engine Technologies
- Human Pose and Action Recognition
- Gear and Bearing Dynamics Analysis
- Vehicle Noise and Vibration Control
- Topic Modeling
- Fuzzy Logic and Control Systems
- Advanced Data Compression Techniques
- High voltage insulation and dielectric phenomena
Nanyang Technological University
2024-2025
State Grid Corporation of China (China)
2024
Beijing Institute of Technology
2013-2023
Civil Aviation Flight University of China
2023
China University of Mining and Technology
2023
Tianjin University
2019-2022
Shenzhen Institute of Information Technology
2021
Shanghai Maritime University
2016-2019
Tsinghua University
2017
Shanghai University of Electric Power
2017
Given the challenge of gathering labeled training data, zero-shot classification, which transfers information from observed classes to recognize unseen classes, has become increasingly popular in computer vision community. Most existing learning methods require a user first provide set semantic visual attributes for each class as side before applying two-step prediction procedure that introduces an intermediate attribute problem. In this paper, we propose novel classification approach...
An echo state network with improved topology (IESN) is proposed for accurate and efficient time series prediction. In this approach, a tighter bound of the property related to Lipshitz constant reservoir activation function maximum structured singular value firstly researched run model at edge chaos. A smooth composite then designed enhance ESN. The exact solved by computing function. Finally, decoupling matrix eigenvalues distributing uniformly in complex plane built as abundant dynamic...
The limited bandwidth of Wi-Fi severely confines the granularity (especially in differentiating multiple subjects) sensing, posing a significant challenge for its wide adoption. Though utilizing channels to expand effective sounds plausible, continuous spectrum stitching towards ultra-wideband (UWB) is far from practical given various constraints (e.g., runtime channel availability and inconsistent responses across bandwidth). To this end, we propose UWB-Fi as novel sensing system with...
Considering the difficulty of selecting sensitive fault features in bearing health diagnosis, a diagnosis method based on complete center frequency distribution feature (CCFDF) is proposed. By making full use sensitivity to signal spectrum variational mode decomposition (VMD) and extracting under different parameter combinations, CCFDF can effectively characterize difference vibration signals conditions avoid setting problem VMD. Finally, two groups experimental data are used verify...
Cloud detection for ground-based sky images has attracted much attention in cloud-related fields. In this paper, we proposed a cloud algorithm that reduced the sunlight interference image. The solar location method was introduced to track sun image used feature calculation, which suitable case where camera could not be calibrated. Following this, adjustable red green difference (ARGD) using and channels proposed. weight determined by layering region division, classified degree of image,...
Accurate and timely fault diagnosis for the diesel engine is crucial to guarantee it works safely reliably, reduces maintenance costs. A novel method based on variational mode decomposition (VMD) kernel-based fuzzy c-means clustering (KFCM) proposed in this paper. Firstly, VMD algorithm optimized select most suitable K value adaptively. Then KFCM employed classify feature parameters of intrinsic functions (IMFs). Through comparison many different parameters, singular selected finally because...
This paper aims to develop an efficient pattern recognition method for engine fault end-to-end detection based on the echo state network (ESN) and multi-verse optimizer (MVO). Bispectrum is employed transform one-dimensional time-dependent vibration signal into a two-dimensional matrix with more impact features. A sparse input weight-generating algorithm designed ESN. Furthermore, deep ESN model built by fusing fixed convolution kernels autoencoder (AE). novel traveling distance rate (TDR)...
Engine fault detection is critical to enhancing the reliability of modern equipment. However, it challenging obtain a large number high-quality labeled data for engines, which not conducive improving training accuracy deep learning methods. Therefore, this article proposes method combining adaptive recursive variational mode decomposition (ARVMD) and component energy distribution spectrum (CEDS). The paper first introduces into VMD. Then, dynamically selected according power spectral density...
Wi-Fi sensing leveraging plain-text beamforming feedback information (BFI) in multiple-input-multipleoutput (MIMO) systems attracts increasing attention.However, due to the implicit relationship between BFI and channel state (CSI), quantifying capability of poses a challenge building efficient BFI-based algorithms.In this letter, we first derive mathematical model BFI, characterizing its with CSI explicitly, then develop closed-form expression for 2×2 MIMO systems.To enhance efficiency by...
Many modern video coding strategies, such as the H.264/AVC standard, use quadtree-based partition structures for intra macroblocks. Such a structure allows algorithm to adapt complicated and non-stationary nature of natural images. Despite adaptation flexibility quadtree partitions, recent studies have shown that these are not efficient enough (in terms rate-distortion performance) when images can be locally modeled 2D piecewise-smooth signals. These observations motivate us investigate...
An accurate and high-resolution diagnosis enables physical failure analysis (PFA) to identify understand the root-cause of integrated-circuit failure. Despite many existing techniques for improving diagnosis, resolution is still far from ideal, which hinders PFA other analyses. To address this challenge, we extend capability PADRE (physically-aware diagnostic enhancement), a powerful machine learning based improvement technique, with novel, active (AL) selection approach. active-learning (AL...
The Jarzynski equality (JE), which connects the equilibrium free energy with nonequilibrium work statistics, plays a crucial role in quantum thermodynamics. Although practical systems are usually multilevel systems, most tests of JE were executed two-level systems. A rigorous test by directly measuring distribution physical process high-dimensional system remains elusive. Here, we report an experimental single spin-1 system. We realized nondemolition projective measurement this three-level...
Knock is an abnormal combustion phenomenon in gasoline engines. Strong knocks will reduce the efficiency and durability of engine, while with slight engines can run on a high-efficiency state. It necessary to detect knock control state order improve thermal engine. This paper proposes novel approach for detecting engine various intensities based vibration signal block using variational mode decomposition (VMD) semi-supervised local fisher discriminant analysis (SELF). Since quadratic penalty...
Multivariate time series (MTS) analysis and forecasting are crucial in many real-world applications, such as smart traffic management weather forecasting. However, most existing work either focuses on short sequence or makes predictions predominantly with domain features, which is not effective at removing noises irregular frequencies MTS. Therefore, we propose WaveForM, an end-to-end graph enhanced Wavelet learning framework for long FORecasting of WaveForM first utilizes Discrete Transform...
In this study, we investigate the task of data pre-selection, which aims to select instances for labeling from an unlabeled dataset through a single pass, thereby optimizing performance undefined downstream tasks with limited annotation budget. Previous approaches pre-selection relied solely on visual features extracted foundation models, such as CLIP and BLIP-2, but largely ignored powerfulness text features. work, argue that, proper design, joint feature space both vision can yield better...
Detecting out-of-distribution (OOD) data is crucial for ensuring the safe deployment of machine learning models in real-world applications. However, existing OOD detection approaches primarily rely on feature maps or full gradient space information to derive scores neglecting role most important parameters pre-trained network over in-distribution (ID) data. In this study, we propose a novel approach called GradOrth facilitate based one intriguing observation that features identify lie...