A5059492353- Machine Fault Diagnosis Techniques
- Railway Engineering and Dynamics
- Traffic Prediction and Management Techniques
- Railway Systems and Energy Efficiency
- Vehicle License Plate Recognition
- Structural Integrity and Reliability Analysis
- Advanced Measurement and Detection Methods
- Electrical Contact Performance and Analysis
- Vehicle emissions and performance
- Anomaly Detection Techniques and Applications
- Advanced Data and IoT Technologies
- Mineral Processing and Grinding
- Gear and Bearing Dynamics Analysis
- Video Surveillance and Tracking Methods
- Electricity Theft Detection Techniques
- Imbalanced Data Classification Techniques
- Traffic control and management
- Privacy-Preserving Technologies in Data
- Fault Detection and Control Systems
- Vehicle Dynamics and Control Systems
- Fire Detection and Safety Systems
- Fatigue and fracture mechanics
- Occupational Health and Safety Research
Southwest Jiaotong University
2021-2025
Due to device operating environment limitations and data privacy protection, it is frequently difficult obtain sufficient high-quality labeled from devices, resulting in an insufficient generalization ability of fault diagnosis model. Therefore, a high-performance federated learning framework proposed this work, which makes improvements the procedure model aggregation local training. In central server, optimization strategy forgetting Kalman filter (FKF) combined with cubic exponential...
Bogie is the unique connection unit between train body and rails, degenerations of its key elements could seriously threaten safety. In previous works that address fault diagnosis bogie high-speed (HST), only a single railway adopted for modeling, which holds insufficient data characteristics thus leads to poor model generalization ability. this paper, an improved federated learning algorithm proposed, reduces computation costs by wavelet packet decomposition trains local models SecureBoost...
Incomplete data is a prevalent phenomenon in mechanical health management, posing limitations on the development of traditional data-driven strategies for fault diagnosis applications. This paper investigates practical challenge achieving accurate using deep learning algorithms absence historical about composite faults high-speed train bogies. To address this issue, we propose signal processing-assisted method and an attribute description strategy modes based time-domain features. We...
As the preferred mode of travel, high-speed train (HST) and its healthy operation have received extensive attention. In long-term service HST, track irregularity wheel-rail wear may cause all kinds faults towards components bogie, which is only connection between body track. Taking into account unknownness bogie fault during actual operation, it inappropriate to simply convert issue diagnosis problem group classification known faults, as conducted in almost reference works. this paper,...
Due to data security concerns, federated learning (FL) has significant computation and communication costs, which lowers total training effectiveness. This research proposes a new framework, Lightweight FL, resolve this problem by enhancing the current fundamental processes. First, local network comprising numerous lightweight methodologies is designed lower costs of model via small-scale convolution calculation. Second, non-structural pruning fine-tuning performed on premise reduce reducing...
The coupler is an essential component on the train that has function of connecting and buffering. actual dynamic performance directly influences safety comfort vehicle. When heavy haul passes through curve, extreme swing angles couplers will seriously threaten train. Therefore, kernelized correlation filter-template matching (KCF-Match) target tracking algorithm proposed to track position calculate couplers. After tracked area selected, corresponding data are input into KCF model for...
In the field of high-speed train fault diagnosis, great achievements have been made in recent years. Although it has achieved considerable accuracy, most them require a large amount labeled training data, which is difficult to collect and obtain actual industrial environment. To solve this problem, paper proposes diagnosis framework based on federated transfer learning. Under premise protecting data privacy security, create model with excellent generalization capabilities can diagnose client...
The air whistle is utilized for locomotive and railway communication alarm. During the operation of heavy haul train, damage to internal mechanical components caused by body vibration can have an effect on sound whistle. Its upkeep frequently relies manual experience, resulting in inefficient maintenance. Meanwhile, when stowed repair, its high decibel level will cause hearing harm maintenance employees. Therefore, deconstructing characteristic modes signals, this work provides intelligent...