Peng Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Peng Chen has a total of 84 co-authors. Here are the top co-authors by number of shared publications:
- Changbo He
Anhui University
11 shared publications Changbo He
9 shared publications- Yaqiang Jin
School of Qilu Transportation, Shandong University , 17923 Jingshi Road, Lixia District , Jinan, Shandong 250061, China
8 shared publications - Junyu Qi
Reutlingen Research Institute, Reutlingen University, Reutlingen, Germany
6 shared publications Yaqiang Jin
School of Qilu Transportation, Shandong University, Jinan 250061, Shandong, People’s Republic of China
6 shared publications- Chaojun Xu
College of Engineering, Shantou University, Shantou, Guangdong, China
5 shared publications
Latest Publications
- 16 September 2026
- Published by SAGE Publications in Structural Health Monitoring
A cross-domain bearing fault diagnosis method under sample imbalance based on multi-view information cross-integration ensemble learning
Reliable cross-domain bearing fault diagnosis is crucial for rotating machinery health monitoring in complex industrial systems. However, practical applications often face severe sample imbalance and domain distribution shifts, the combination of which frequently causes the model to be biased toward majority classes. To overcome these challenges, we propose multi-view information cross-integration ensemble learning (MICE). Firstly, MICE employs a multi-view cross-integration strategy to generate mixed signals, thereby extracting complementary feature representations while reducing redundancy. Secondly, to handle sample imbalance, a consistency voting entropy optimization strategy analyzes prediction scores to set dynamic thresholds and generate high-confidence pseudo-labels. This process enables the model to effectively capture critical minority-class features. Finally, a dual-path collaborative learning mechanism integrates inter-view feature transfer with inter-domain alignment, and employs a confusion-class focusing strategy to explicitly distinguish confusable categories. Extensive experiments on two public datasets validate the superiority and strong cross-domain generalization capability of the proposed MICE method.
- 16 September 2026
- Published by SAGE Publications in Structural Health Monitoring
A cross-domain bearing fault diagnosis method under sample imbalance based on multi-view information cross-integration ensemble learning
Reliable cross-domain bearing fault diagnosis is crucial for rotating machinery health monitoring in complex industrial systems. However, practical applications often face severe sample imbalance and domain distribution shifts, the combination of which frequently causes the model to be biased toward majority classes. To overcome these challenges, we propose multi-view information cross-integration ensemble learning (MICE). Firstly, MICE employs a multi-view cross-integration strategy to generate mixed signals, thereby extracting complementary feature representations while reducing redundancy. Secondly, to handle sample imbalance, a consistency voting entropy optimization strategy analyzes prediction scores to set dynamic thresholds and generate high-confidence pseudo-labels. This process enables the model to effectively capture critical minority-class features. Finally, a dual-path collaborative learning mechanism integrates inter-view feature transfer with inter-domain alignment, and employs a confusion-class focusing strategy to explicitly distinguish confusable categories. Extensive experiments on two public datasets validate the superiority and strong cross-domain generalization capability of the proposed MICE method.
- 29 May 2026
- Published by IOP Publishing in Measurement Science and Technology
Frequency-consistency regularization in dual-stream time–frequency learning for semi-supervised rotating machinery fault diagnosis
Semi-supervised learning presents a promising avenue for gearbox fault diagnosis under label-scarce conditions, yet it remains hampered by the unreliability of pseudo-labels and the spectral distortion induced by generic time-domain augmentations. These deficiencies frequently precipitate error accumulation during self-training, thereby degrading diagnostic robustness in complex industrial environments. To address these challenges, this paper proposes the Frequency-Consistency Regularization in Dual-Stream Time-Frequency Learning (FCR-DSTFL) framework. We introduce a physics-informed Multi-Scale Frequency Augmentation (MSFA) strategy that utilizes a kurtosis-driven adaptive reconstruction to generate semantically faithful spectral views, effectively solving the cross-modality incompatibility of standard perturbations. Furthermore, we develop a dual-stream frequency-consistency mechanism that suppresses "Spectral Pseudo-Label Drift" by enforcing predictive alignment between raw temporal signals and their frequency-refined counterparts, ensuring that only high-fidelity supervision signals are utilized. Experimental results on a parallel gearbox system and a Drivetrain Prognostics Simulator demonstrate that the proposed method achieves superior diagnostic precision, outperforming state-of-the-art semi-supervised baselines by significant margins in low-data regimes.
- 29 May 2026
- Published by IOP Publishing in Measurement Science and Technology
Frequency-consistency regularization in dual-stream time–frequency learning for semi-supervised rotating machinery fault diagnosis
Semi-supervised learning presents a promising avenue for gearbox fault diagnosis under label-scarce conditions, yet it remains hampered by the unreliability of pseudo-labels and the spectral distortion induced by generic time-domain augmentations. These deficiencies frequently precipitate error accumulation during self-training, thereby degrading diagnostic robustness in complex industrial environments. To address these challenges, this paper proposes the Frequency-Consistency Regularization in Dual-Stream Time-Frequency Learning (FCR-DSTFL) framework. We introduce a physics-informed Multi-Scale Frequency Augmentation (MSFA) strategy that utilizes a kurtosis-driven adaptive reconstruction to generate semantically faithful spectral views, effectively solving the cross-modality incompatibility of standard perturbations. Furthermore, we develop a dual-stream frequency-consistency mechanism that suppresses "Spectral Pseudo-Label Drift" by enforcing predictive alignment between raw temporal signals and their frequency-refined counterparts, ensuring that only high-fidelity supervision signals are utilized. Experimental results on a parallel gearbox system and a Drivetrain Prognostics Simulator demonstrate that the proposed method achieves superior diagnostic precision, outperforming state-of-the-art semi-supervised baselines by significant margins in low-data regimes.
- Open Access
- 1 May 2026
- Published by KEAI Publishing in Chinese Journal of Mechanical Engineering
TimeGPT for Mechanical System Degradation Assessment: A Large Time Series Model Approach
- 1 May 2026
- Published by KEAI Publishing in Chinese Journal of Mechanical Engineering
Open AccessTimeGPT for Mechanical System Degradation Assessment: A Large Time Series Model Approach
- 5 March 2026
- Published by ASME International in ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part B. Mechanical Engineering
- Vol. 12 (4)
CyclicalNet: A Dense-Connected Architecture for Multiscale Cyclic Feature Learning in Planetary Gearbox Fault Diagnosis
Planetary gearboxes, crucial for industrial applications like wind turbines, automobiles, and robotics, are susceptible to failures due to complex vibrations and harsh conditions, making the development of effective fault diagnosis methods essential. While general-purpose deep learning models—such as Transformers, ResNets, DenseNets, and generative adversarial networks (GANs)—have shown promise in identifying intricate patterns within time series data for fault diagnosis, conventional one-dimensional convolutional kernels have inherent limitations, as they can only capture features from adjacent time series points. Additionally, methods that utilize attention mechanisms often struggle to uncover consistent dependencies among sparse data points that may be concealed within complex patterns. This paper introduces CyclicalNet, an innovative deep learning architecture designed to address these fundamental limitations in planetary gearbox fault diagnosis. The architecture uniquely integrates cyclical blocks, encoder blocks, and dense connections to process high amplitude–frequency components, traditionally overlooked in conventional models. A strategic reshape operation transforms one-dimensional sequential data into two-dimensional representations, enabling more structured and efficient feature extraction while facilitating simultaneous modeling of temporal and periodic features. The incorporation of dense connections promotes multiscale feature learning and enhances information flow across the network, resulting in richer and more discriminative feature representations. Through comprehensive experimental evaluation on planetary gearbox datasets, CyclicalNet demonstrates superior performance in fault diagnosis accuracy compared to existing methods, establishing a new benchmark in predictive maintenance technology.
- 5 March 2026
- Published by ASME International in ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part B. Mechanical Engineering
- Vol. 12 (4)
CyclicalNet: A Dense-Connected Architecture for Multiscale Cyclic Feature Learning in Planetary Gearbox Fault Diagnosis
Planetary gearboxes, crucial for industrial applications like wind turbines, automobiles, and robotics, are susceptible to failures due to complex vibrations and harsh conditions, making the development of effective fault diagnosis methods essential. While general-purpose deep learning models—such as Transformers, ResNets, DenseNets, and generative adversarial networks (GANs)—have shown promise in identifying intricate patterns within time series data for fault diagnosis, conventional one-dimensional convolutional kernels have inherent limitations, as they can only capture features from adjacent time series points. Additionally, methods that utilize attention mechanisms often struggle to uncover consistent dependencies among sparse data points that may be concealed within complex patterns. This paper introduces CyclicalNet, an innovative deep learning architecture designed to address these fundamental limitations in planetary gearbox fault diagnosis. The architecture uniquely integrates cyclical blocks, encoder blocks, and dense connections to process high amplitude–frequency components, traditionally overlooked in conventional models. A strategic reshape operation transforms one-dimensional sequential data into two-dimensional representations, enabling more structured and efficient feature extraction while facilitating simultaneous modeling of temporal and periodic features. The incorporation of dense connections promotes multiscale feature learning and enhances information flow across the network, resulting in richer and more discriminative feature representations. Through comprehensive experimental evaluation on planetary gearbox datasets, CyclicalNet demonstrates superior performance in fault diagnosis accuracy compared to existing methods, establishing a new benchmark in predictive maintenance technology.
- 5 March 2026
- Published by ASME International in ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part B. Mechanical Engineering
- Vol. 12 (4)
EKBD-MK: Entropy-Kurtosis Bilateral Discernment With Maximum Kurtosis Blind Deconvolution for Fault Diagnosis in Wind Turbine Systems
Wind turbines, as critical components of renewable energy systems, face significant challenges in maintaining operational reliability due to their bearing systems operating under challenging mechanical and environmental conditions. These conditions are further complicated by transient events generating sudden mechanical impulses and persistent harmonic noise from mechanical resonance, which obscure early indicators of component degradation and hinder fault detection. To address these challenges, this research proposes entropy-kurtosis bilateral discernment with maximum kurtosis blind deconvolution (EKBD-MK), a novel diagnostic framework that systematically suppresses nonstationary transient disturbances and irrelevant harmonic components in vibration signals. By employing a Shannon entropy-based noise quantification approach alongside physical interpretations of center frequencies, the proposed method ensures robustness in complex noise environments. Additionally, a signal decomposition strategy through variational mode decomposition (VMD) is developed with an entropy-frequency dual-constraint system to optimize fault information concentration, prevent feature loss, and maintain signal integrity. To further enhance diagnostic accuracy, an adaptive blind deconvolution (BD) strategy incorporating multicomponent preservation and spectral kurtosis (SK)-based enhancement of fault features is implemented, ensuring parameter insensitivity and effective noise suppression. Experimental validation through real case study demonstrates the efficacy of this framework, significantly improving diagnostic precision compared to current methods.
- 5 March 2026
- Published by ASME International in ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part B. Mechanical Engineering
- Vol. 12 (4)
EKBD-MK: Entropy-Kurtosis Bilateral Discernment With Maximum Kurtosis Blind Deconvolution for Fault Diagnosis in Wind Turbine Systems
Wind turbines, as critical components of renewable energy systems, face significant challenges in maintaining operational reliability due to their bearing systems operating under challenging mechanical and environmental conditions. These conditions are further complicated by transient events generating sudden mechanical impulses and persistent harmonic noise from mechanical resonance, which obscure early indicators of component degradation and hinder fault detection. To address these challenges, this research proposes entropy-kurtosis bilateral discernment with maximum kurtosis blind deconvolution (EKBD-MK), a novel diagnostic framework that systematically suppresses nonstationary transient disturbances and irrelevant harmonic components in vibration signals. By employing a Shannon entropy-based noise quantification approach alongside physical interpretations of center frequencies, the proposed method ensures robustness in complex noise environments. Additionally, a signal decomposition strategy through variational mode decomposition (VMD) is developed with an entropy-frequency dual-constraint system to optimize fault information concentration, prevent feature loss, and maintain signal integrity. To further enhance diagnostic accuracy, an adaptive blind deconvolution (BD) strategy incorporating multicomponent preservation and spectral kurtosis (SK)-based enhancement of fault features is implemented, ensuring parameter insensitivity and effective noise suppression. Experimental validation through real case study demonstrates the efficacy of this framework, significantly improving diagnostic precision compared to current methods.
- 24 February 2026
- Published by Springer Nature in Nonlinear Dynamics
- Vol. 114 (4) , 244
Markovian spectral transition modeling with temporal dependencies for railway bogie axle bearing diagnostics in non-stationary transient environments
- 24 February 2026
- Published by Springer Nature in Nonlinear Dynamics
- Vol. 114 (4) , 244
Markovian spectral transition modeling with temporal dependencies for railway bogie axle bearing diagnostics in non-stationary transient environments
- 14 February 2026
- Published by Elsevier in Measurement
- Vol. 270
Cross-correlation heat-maps synthesis with Walsh–Fourier transformation in articulated robotic systems diagnostics
- 14 February 2026
- Published by Elsevier in Measurement
- Vol. 270
Cross-correlation heat-maps synthesis with Walsh–Fourier transformation in articulated robotic systems diagnostics
- 4 February 2026
- Published by SAGE Publications in Structural Health Monitoring
Semi-Supervised Transfer Graph Representation Learning with Few-Shot Adaptation for gearbox diagnostics under extraneous transient noise
Gearboxes are critical mechanical components widely deployed in industrial applications, where their reliable operation directly impacts system safety and efficiency. However, conventional fault diagnostic approaches face significant challenges when operating under extraneous transient noise conditions, particularly with limited labeled fault samples. These challenges manifest as performance degradation with extremely sparse labeled datasets, vulnerability in pseudo-label generation mechanisms under intense transient noise, and inconsistent feature scale representations due to noise-induced interference. Furthermore, existing methods struggle to maintain diagnostic accuracy when confronted with both data scarcity and transient disturbances, often resulting in compromised model generalization and unreliable fault classification. To address these limitations, this research proposes the Semi-Supervised Transfer Graph Representation Learning with Few-Shot Adaptation (SSTGRL-FSA) framework, featuring three innovative components: a novel pseudo-label reliability enhancement mechanism leveraging systematic knowledge transfer from established source domains, an advanced label transmission and matching strategy exploiting homologous signal patterns across operational domains, and an integrated first-order Markov state probability transition matrix with amplitude-constrained scaling. SSTGRL-FSA significantly advances the field by effectively handling both labeled data scarcity and transient noise interference while enhancing model robustness through sophisticated temporal dependency modeling and stable feature scale maintenance, ultimately providing a more reliable and practical solution for industrial gearbox fault diagnosis under challenging operational conditions.
- 4 February 2026
- Published by SAGE Publications in Structural Health Monitoring
Semi-Supervised Transfer Graph Representation Learning with Few-Shot Adaptation for gearbox diagnostics under extraneous transient noise
Gearboxes are critical mechanical components widely deployed in industrial applications, where their reliable operation directly impacts system safety and efficiency. However, conventional fault diagnostic approaches face significant challenges when operating under extraneous transient noise conditions, particularly with limited labeled fault samples. These challenges manifest as performance degradation with extremely sparse labeled datasets, vulnerability in pseudo-label generation mechanisms under intense transient noise, and inconsistent feature scale representations due to noise-induced interference. Furthermore, existing methods struggle to maintain diagnostic accuracy when confronted with both data scarcity and transient disturbances, often resulting in compromised model generalization and unreliable fault classification. To address these limitations, this research proposes the Semi-Supervised Transfer Graph Representation Learning with Few-Shot Adaptation (SSTGRL-FSA) framework, featuring three innovative components: a novel pseudo-label reliability enhancement mechanism leveraging systematic knowledge transfer from established source domains, an advanced label transmission and matching strategy exploiting homologous signal patterns across operational domains, and an integrated first-order Markov state probability transition matrix with amplitude-constrained scaling. SSTGRL-FSA significantly advances the field by effectively handling both labeled data scarcity and transient noise interference while enhancing model robustness through sophisticated temporal dependency modeling and stable feature scale maintenance, ultimately providing a more reliable and practical solution for industrial gearbox fault diagnosis under challenging operational conditions.