An unsupervised domain adaptation method with coral-maximum mean discrepancy for subway train transmission system fault diagnosis

Smart and Resilient Transport, Journal Year: 2025, Volume and Issue: unknown

Published: March 28, 2025

Purpose In modern railway train systems, the safety of fault diagnosis technology is crucial. Currently, data-driven mainstream algorithms have achieved good results in diagnosis. However, current research typically focuses on equipment data set itself, with training and test sets often being same distribution. Due to diversity operating conditions, applying these methods industrial practice can be challenging. The purpose this study address challenges practice, given conditions. Design/methodology/approach To issue, authors propose an unsupervised domain adaptation method for subway transmission which combines convolutional neural networks (CNN) a novel algorithm, coral-maximum mean discrepancy (Cor-MMD). Findings First, feature extraction performed systems under various are divided into training, validation sets. Next, CNN trained using obtain model weight files. Then, based files along labeled unlabeled set, deep optimized by minimizing Cor-MMD loss. Originality/value Finally, accuracy was tested effectiveness our verified through comparative experiments.

Language: Английский

---

MSC-APFNN: Fault Diagnosis Deep Transfer Network for Rotating Machine based on Multi-scale Convolutional Extraction and Adaptive Pruning Fuzzy Inference System

Lecture notes in electrical engineering, Journal Year: 2025, Volume and Issue: unknown, P. 310 - 324

Published: Jan. 1, 2025

Language: Английский

---

Intelligent Fault Diagnosis Method of Train Axle Box Bearing Based on Improved Deep Reinforcement Learning

Lecture notes in electrical engineering, Journal Year: 2025, Volume and Issue: unknown, P. 430 - 440

Published: Jan. 1, 2025

Language: Английский

---

Unsupervised fault detection with multi-source anomaly sensitivity enhancing convolutional autoencoder for high-speed train bogie bearings

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 127570 - 127570

Published: April 1, 2025

Language: Английский

---

Intelligent Fault Diagnosis for CNC Through the Integration of Large Language Models and Domain Knowledge Graphs

Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: April 1, 2025

Language: Английский

---

Condition monitoring of train transmission systems based on multimodal fusion improved transformer network

Sound&Vibration, Journal Year: 2025, Volume and Issue: 59(2), P. 2904 - 2904

Published: April 25, 2025

The train transmission system is a critical component of railway operations, playing pivotal role in ensuring service safety and reliability. However, existing condition monitoring approaches face two major challenges: (1) the coupling rich multimodal signals, such as vibration, acoustics, current, rotational speed, often overlooked, limiting accuracy; (2) small data problem signals adversely affects performance neural networks. To address these issues, this paper proposes Multimodal Fusion Improved Transformer Network for Condition Monitoring Train Transmission Systems. proposed network first explores interdependencies among different modalities compresses to reduced dimensions through correlation analysis. It then infers global dependencies computing self-attention scores based on Q, K, V matrices. approach better than traditional CNN-based models handling single-modality constraints, with former demonstrated be more accurate trustworthy publicly available datasets.

Language: Английский

---

A new multi-layer adaptation cross-domain model for bearing fault diagnosis under different operating conditions

Measurement Science and Technology, Journal Year: 2024, Volume and Issue: 35(10), P. 106116 - 106116

Published: July 17, 2024

Abstract Bearing faults under different operating conditions often cannot be diagnosed by models trained a single operational condition. Additionally, the extraction of domain-invariant features in domain adaptation (DA) algorithms is also challenge. To address aforementioned issues, multi-layer model based on an improved sparse autoencoders (SAEs) and dual-domain distance mechanism (ISAE-DDM) proposed. First, feature capability traditional SAEs enhanced strategy that combines mean squared error with absolute error. Subsequently, data multiple hidden layers are extracted ISAE. Then, distribution discrepancy between source target measured approach Wasserstein multi-kernel maximum discrepancy. loss each layer assigned weighting parameters. Finally, joint metric DA across constructed to achieve better alignment. The performance proposed method demonstrated two bearing experiments. Moreover, this exhibits higher stability, generalization capabilities compared other methods.

Language: Английский

---

A Fault Detection of Aero-engine Rolling Bearings based on CNN-BiLSTM Network Integrated Cross-Attention

Measurement Science and Technology, Journal Year: 2024, Volume and Issue: 35(12), P. 126116 - 126116

Published: Sept. 2, 2024

Abstract Aero-engine rolling bearings are essential for engine health, in which disruptive failures can be prevented and reduce great losses air flight. To improve the efficiency of fault detection, an improved network, named CNN- BiLSTM -Cross-Attention (CBLCA) was proposed. The Bidirectional Long Short-Term Memory (BiLSTM) layer captures temporal features as input data. cross-attention mechanism is integrated with Convolutional neural networks (CNN) respectively. More important feature information identified CBLCA model. proposed model also validated open-sourced aero-engine data set. identification accuracy, a novel method that combines fast Fourier transform Variational mode decomposition used preprocessing. Each original signal sample transformed into set containing richer information, number significantly increased entire dataset. Compared some existing LSTM models, such LSTM, BiLSTM, CNN-BiLSTM, CNN-LSTM, classification accuracy by 55%, 54%, 5%, 7%, processing vibration signals reliability diagnosis bearings.

Language: Английский

---

IF-EDAAN: An information fusion-enhanced domain adaptation attention network for unsupervised transfer fault diagnosis

Mechanical Systems and Signal Processing, Journal Year: 2024, Volume and Issue: 224, P. 112180 - 112180

Published: Nov. 26, 2024

Language: Английский

---

Few-shot sample multi-class incremental fault diagnosis for gearbox based on convolutional-attention fusion network

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: unknown, P. 125918 - 125918

Published: Nov. 1, 2024

Language: Английский

---