Keting Hu
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
Keting Hu has a total of 40 co-authors. Here are the top co-authors by number of shared publications:
- Zhigang Liu
Southwest Jiaotong University
15 shared publications - Haonan Yang
School of Electrical Engineering, Southwest Jiaotong University, Chengdu 610097, China
4 shared publications - Haoyu Wang
Southwest Jiaotong University
4 shared publications - Ibrahim Adamu Tasiu
School of Electronic Information Engineering, China West Normal University, Nanchong 637002, China
4 shared publications - Yizhuo Li
Southwest Jiaotong University
3 shared publications - Francesco Iannuzzo
Department of Energy, Politecnico di Torino, Turin, Italy
2 shared publications
Latest Publications
- 17 June 2026
- Published by Springer Nature
- p. 909-921
Fault Diagnosis of PV Inverter Open-Switch Fault with Low Sampling Frequency Data
- 17 June 2026
- Published by Springer Nature
- p. 909-921
Fault Diagnosis of PV Inverter Open-Switch Fault with Low Sampling Frequency Data
- Top Cited Papers
- 4 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. PP (5) , 1
BCLIP-ADer: A Bayesian Prompt Contrastive Language-Image Pretraining Method for Catenary Component Anomaly Detection in Electrified Railways
As an essential subsystem of electrified railway operation and maintenance, intelligent detection of catenary support components still faces several critical challenges: (1) the number of abnormal (negative) samples for components is severely limited; (2) component anomalies are highly diverse and exhibit heterogeneous visual characteristics; and (3) existing models generally show unsatisfactory detection performance when confronted with previously unseen anomaly types. To address these issues, this paper proposes a novel few-shot anomaly detection model for catenary components, termed BCLIP-ADer, built upon a Bayesian prompt contrastive vision–language pretraining framework. Specifically, a Bayesian prompt flow module (PFM) is designed to regularize the text prompt space via the jointly learned image-specific feature distribution (ISFD) and image-agnostic feature distribution (IAFD), thereby mitigating the degradation in detection performance on unseen component anomalies. Monte Carlo sampling over these learned distributions is further employed to generate diverse text prompts, leading to more comprehensive coverage of the prompt space. In addition, a cross-modal feature refinement module (CFRM) is designed to more effectively align dynamic text embeddings with fine-grained image features, thus enhancing anomaly detection at the component level. Finally, extensive experiments conducted on a UAV-based catenary dataset (CSCUD) demonstrate the effectiveness and superiority of the proposed approach. Specifically, the proposed method achieves I-AUROC/I-AP/I-F1_max scores of 94.2/93.2/93.1 under few-shot conditions.
- 4 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. PP (5) , 1
Top Cited PapersBCLIP-ADer: A Bayesian Prompt Contrastive Language-Image Pretraining Method for Catenary Component Anomaly Detection in Electrified Railways
As an essential subsystem of electrified railway operation and maintenance, intelligent detection of catenary support components still faces several critical challenges: (1) the number of abnormal (negative) samples for components is severely limited; (2) component anomalies are highly diverse and exhibit heterogeneous visual characteristics; and (3) existing models generally show unsatisfactory detection performance when confronted with previously unseen anomaly types. To address these issues, this paper proposes a novel few-shot anomaly detection model for catenary components, termed BCLIP-ADer, built upon a Bayesian prompt contrastive vision–language pretraining framework. Specifically, a Bayesian prompt flow module (PFM) is designed to regularize the text prompt space via the jointly learned image-specific feature distribution (ISFD) and image-agnostic feature distribution (IAFD), thereby mitigating the degradation in detection performance on unseen component anomalies. Monte Carlo sampling over these learned distributions is further employed to generate diverse text prompts, leading to more comprehensive coverage of the prompt space. In addition, a cross-modal feature refinement module (CFRM) is designed to more effectively align dynamic text embeddings with fine-grained image features, thus enhancing anomaly detection at the component level. Finally, extensive experiments conducted on a UAV-based catenary dataset (CSCUD) demonstrate the effectiveness and superiority of the proposed approach. Specifically, the proposed method achieves I-AUROC/I-AP/I-F1_max scores of 94.2/93.2/93.1 under few-shot conditions.
- 3 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-13
A Model–Data Hybrid Fault Diagnosis Approach With Few-Shot Learning for PV-Grid-Connected Three-Level ANPC Inverter
Intelligent monitoring and fault diagnosis are essential for ensuring the reliable operation of PV grid-connected three-level Active Neutral-Point-Clamped (3L-ANPC) inverters. However, under complex climatic conditions, obtaining sufficient fault samples is extremely challenging, and the diversity of fault types further increases the difficulty for traditional methods. To address this unique challenge, this study proposes a model–data hybrid approach for open-switch fault diagnosis. First, to construct robust fault diagnostic variables, a model-based sliding mode observer is constructed to extract residual currents and phase diagnostic variables as input features. Then, to achieve a comprehensive representation of fault characteristics, a Group-Hybrid Former is designed to capture intra-group feature correlations, and integrated with a Siamese network to enable efficient feature learning under few-shot conditions. Experimental results show that the proposed method achieves a diagnostic accuracy of 97.79% in few-shot scenarios, outperforming state-of-the-art deep learning and few-shot methods by 21.75%, thereby demonstrating its robustness and reliability.
- 3 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-13
A Model–Data Hybrid Fault Diagnosis Approach With Few-Shot Learning for PV-Grid-Connected Three-Level ANPC Inverter
Intelligent monitoring and fault diagnosis are essential for ensuring the reliable operation of PV grid-connected three-level Active Neutral-Point-Clamped (3L-ANPC) inverters. However, under complex climatic conditions, obtaining sufficient fault samples is extremely challenging, and the diversity of fault types further increases the difficulty for traditional methods. To address this unique challenge, this study proposes a model–data hybrid approach for open-switch fault diagnosis. First, to construct robust fault diagnostic variables, a model-based sliding mode observer is constructed to extract residual currents and phase diagnostic variables as input features. Then, to achieve a comprehensive representation of fault characteristics, a Group-Hybrid Former is designed to capture intra-group feature correlations, and integrated with a Siamese network to enable efficient feature learning under few-shot conditions. Experimental results show that the proposed method achieves a diagnostic accuracy of 97.79% in few-shot scenarios, outperforming state-of-the-art deep learning and few-shot methods by 21.75%, thereby demonstrating its robustness and reliability.
- 13 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
Global Feature Perception Network for Open-Switch Fault Diagnosis in Grid-Connected Photovoltaic Inverters Under Imbalanced Data Conditions
Intelligent monitoring and fault diagnosis of photovoltaic (PV) grid-connected inverters are important. However, the accuracy of traditional fault diagnosis methods is challenged due to multiple data imbalance. To address this unique challenge, this study proposes the Global Feature Perception Network (GFPN) framework. First, to effectively capture the operational characteristics of PV inverters under complex conditions, a dilated feature boost block is designed to improve multi-level feature representation. Second, to ensure accurate judgments across various operating scenarios, a dilated feature mix transformer is designed to capture both causal and non-causal relationships between different features. Additionally, to enhance diagnostic performance for minority fault samples, the enhanced dual focal loss is proposed, improving the fault diagnosis effectiveness of model under imbalanced conditions. Experimental results show that even under the most challenging data imbalance scenarios, GFPN achieves fault diagnosis accuracies of 96.09% and 96.25% on PV grid-connected inverters and the Hardware-in-the-Loop platform, respectively.
- 13 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
Global Feature Perception Network for Open-Switch Fault Diagnosis in Grid-Connected Photovoltaic Inverters Under Imbalanced Data Conditions
Intelligent monitoring and fault diagnosis of photovoltaic (PV) grid-connected inverters are important. However, the accuracy of traditional fault diagnosis methods is challenged due to multiple data imbalance. To address this unique challenge, this study proposes the Global Feature Perception Network (GFPN) framework. First, to effectively capture the operational characteristics of PV inverters under complex conditions, a dilated feature boost block is designed to improve multi-level feature representation. Second, to ensure accurate judgments across various operating scenarios, a dilated feature mix transformer is designed to capture both causal and non-causal relationships between different features. Additionally, to enhance diagnostic performance for minority fault samples, the enhanced dual focal loss is proposed, improving the fault diagnosis effectiveness of model under imbalanced conditions. Experimental results show that even under the most challenging data imbalance scenarios, GFPN achieves fault diagnosis accuracies of 96.09% and 96.25% on PV grid-connected inverters and the Hardware-in-the-Loop platform, respectively.
- 22 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 73 (6) , 9009-9014
Super-Twisting-Observer-Based Equivalent-Input-Disturbance Compensation Suspension Control of Medium-Low Speed Maglev
This paper presents an easily implementable disturbance compensation suspension control method for the medium-low speed maglev suspension system. The impact of state observation accuracy on EID estimation is firstly examined, suggesting that accurate state observation is crucial for estimating EID. Building upon this foundation, a super-twisting-observer-based EID (STO-EID) estimation method is proposed by incorporating the super-twisting observer (STO) and EID estimation. The stability of both the STO and the overall control system is demonstrated using Lyapunov's method. Finally, simulations and experiments are performed to assess the performance and effectiveness of the proposed STO-EID method in comparison to the conventional sliding mode observer EID method and the linear quadratic Gaussian method, showing superior results.
- 22 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 73 (6) , 9009-9014
Super-Twisting-Observer-Based Equivalent-Input-Disturbance Compensation Suspension Control of Medium-Low Speed Maglev
This paper presents an easily implementable disturbance compensation suspension control method for the medium-low speed maglev suspension system. The impact of state observation accuracy on EID estimation is firstly examined, suggesting that accurate state observation is crucial for estimating EID. Building upon this foundation, a super-twisting-observer-based EID (STO-EID) estimation method is proposed by incorporating the super-twisting observer (STO) and EID estimation. The stability of both the STO and the overall control system is demonstrated using Lyapunov's method. Finally, simulations and experiments are performed to assess the performance and effectiveness of the proposed STO-EID method in comparison to the conventional sliding mode observer EID method and the linear quadratic Gaussian method, showing superior results.
- 24 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. 8 (4) , 4595-4606
Flux-Based Open-Switch Fault Diagnosis and Fault Tolerance for IM Drives With Predictive Torque/Flux Control
In this article, open-switch fault diagnosis and fault tolerance are proposed for induction machine (IM) drives with predictive torque control (PTC) or predictive flux control (PFC). Different from the conventional diagnosis methods for drives with finite control set model predictive control (FCS-MPC), which are based on the current, the proposed method is based on the flux error. As a result, this method can be extended to IM drive systems with PTC or PFC. It detects the fault by computing and accumulating the error between the predicted flux from the last sampling interval and the estimated flux at the current interval. Then, the polarity of the accumulated residual is utilized to localize the fault. For the fault-tolerant method, the three-phase four-switch (TPFS) inverter is chosen as it can keep working with only four switches. To reduce the torque ripple caused by the TPFS topology, a three-vector PFC method is proposed. Hardware-in-the-loop (HIL) tests are carried out to validate the effectiveness of the proposed methods. Compared to conventional diagnosis methods, the proposed one has a much faster diagnosis speed and the three-vector PFC outperforms the one- and two-vector methods.
- 24 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. 8 (4) , 4595-4606
Flux-Based Open-Switch Fault Diagnosis and Fault Tolerance for IM Drives With Predictive Torque/Flux Control
In this article, open-switch fault diagnosis and fault tolerance are proposed for induction machine (IM) drives with predictive torque control (PTC) or predictive flux control (PFC). Different from the conventional diagnosis methods for drives with finite control set model predictive control (FCS-MPC), which are based on the current, the proposed method is based on the flux error. As a result, this method can be extended to IM drive systems with PTC or PFC. It detects the fault by computing and accumulating the error between the predicted flux from the last sampling interval and the estimated flux at the current interval. Then, the polarity of the accumulated residual is utilized to localize the fault. For the fault-tolerant method, the three-phase four-switch (TPFS) inverter is chosen as it can keep working with only four switches. To reduce the torque ripple caused by the TPFS topology, a three-vector PFC method is proposed. Hardware-in-the-loop (HIL) tests are carried out to validate the effectiveness of the proposed methods. Compared to conventional diagnosis methods, the proposed one has a much faster diagnosis speed and the three-vector PFC outperforms the one- and two-vector methods.
- 7 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. 8 (3) , 3848-3858
A Multi-Frequency Input-Admittance Model of Locomotive Rectifier Considering PWM Sideband Harmonic Coupling in Electrical Railways
Electrical railway harmonic instability issues are common in the high-frequency range. The effective frequency of the traditional converter’s small-signal averaging model is below 1/2 switching frequency since the pulse width modulation (PWM) sideband harmonic components are ignored. In this article, the dynamic propagations of perturbation frequency and the generated PWM sideband components are constructed first. Then the locomotive rectifier’s multi-frequency input-admittance model is derived appropriately. Afterward, an admittance conversion approach is used to convert the multi-frequency model into the single-input–single-output (SISO) model whereas retaining the sideband frequency couplings. The proposed SISO model is more accurate than the traditional small-signal averaging model in the frequency range higher than 1/2 switching frequency. It is found that PWM sideband harmonics dominate the locomotive rectifier’s input-admittance characteristic higher than 1/2 switching frequency. Finally, based on the proposed model, the influence of different switching frequencies, control bandwidths, and traction network impedance on system harmonic stability is revealed by the hardware-in-the-loop (HIL) results.
- 7 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. 8 (3) , 3848-3858
A Multi-Frequency Input-Admittance Model of Locomotive Rectifier Considering PWM Sideband Harmonic Coupling in Electrical Railways
Electrical railway harmonic instability issues are common in the high-frequency range. The effective frequency of the traditional converter’s small-signal averaging model is below 1/2 switching frequency since the pulse width modulation (PWM) sideband harmonic components are ignored. In this article, the dynamic propagations of perturbation frequency and the generated PWM sideband components are constructed first. Then the locomotive rectifier’s multi-frequency input-admittance model is derived appropriately. Afterward, an admittance conversion approach is used to convert the multi-frequency model into the single-input–single-output (SISO) model whereas retaining the sideband frequency couplings. The proposed SISO model is more accurate than the traditional small-signal averaging model in the frequency range higher than 1/2 switching frequency. It is found that PWM sideband harmonics dominate the locomotive rectifier’s input-admittance characteristic higher than 1/2 switching frequency. Finally, based on the proposed model, the influence of different switching frequencies, control bandwidths, and traction network impedance on system harmonic stability is revealed by the hardware-in-the-loop (HIL) results.
- 28 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 70 (1) , 303-318
Fuzzy Observer-Based Control for the Traction Dual Rectifiers in High-Speed Train
This paper proposes a fuzzy-observer-based control algorithm to improve the efficiency of traction dual rectifiers of high-speed trains and address the low-frequency oscillation issue induced by the interaction of the DC-link voltage controller with the traction dual rectifiers circuit parameters. The proposed observer checks for the errors in the system model by estimating the system premise variables, and the fuzzy controller then accurately compensates the estimated variables. The method is deduced from the Takagi-Sugeno fuzzy logic model, which is described by a class of IF-THEN fuzzy systems. Fast computation with high precision is what makes them superior to proportional-integral-derivatives controllers. Their capability to deal with nonlinearity in time-varying systems makes them a suitable choice for traction dual rectifier control. The simulation model is formed in SIMULINK, and the waveforms are checked with a conventional fuzzy controller, predictive control, and conventional proportional-integral control. The proposed method can achieve better performance, such as fast transient, more damping, and small overshoot. Furthermore, the method can maintain stability when the system is subjected to external disturbance. In the end, hardware-in-the-loop experiments are performed to validate the proposed control algorithm.
- 28 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 70 (1) , 303-318
Fuzzy Observer-Based Control for the Traction Dual Rectifiers in High-Speed Train
This paper proposes a fuzzy-observer-based control algorithm to improve the efficiency of traction dual rectifiers of high-speed trains and address the low-frequency oscillation issue induced by the interaction of the DC-link voltage controller with the traction dual rectifiers circuit parameters. The proposed observer checks for the errors in the system model by estimating the system premise variables, and the fuzzy controller then accurately compensates the estimated variables. The method is deduced from the Takagi-Sugeno fuzzy logic model, which is described by a class of IF-THEN fuzzy systems. Fast computation with high precision is what makes them superior to proportional-integral-derivatives controllers. Their capability to deal with nonlinearity in time-varying systems makes them a suitable choice for traction dual rectifier control. The simulation model is formed in SIMULINK, and the waveforms are checked with a conventional fuzzy controller, predictive control, and conventional proportional-integral control. The proposed method can achieve better performance, such as fast transient, more damping, and small overshoot. Furthermore, the method can maintain stability when the system is subjected to external disturbance. In the end, hardware-in-the-loop experiments are performed to validate the proposed control algorithm.
- 25 November 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. 7 (3) , 1031-1046
A Model Predictive Control Considering Parameters and System Uncertainties for Suppressing Low-Frequency Oscillations of Traction Dual Rectifiers
Low-frequency oscillation (LFO) overvoltage problems often occur in vehicle-grid coupled systems of high-speed railways. In order to suppress LFO and make up for the problem of weak control effect caused by the large deviation between the model predictive control (MPC) model and the actual system, the uncertainties in parameters and system are enough considered and a combination of MPC and deviation error correction (DEC) is proposed in this paper. First, the error caused by the parameter mismatch, discretization, and sampling process from system noise is analyzed. Then, the DEC is designed to compensate for the calculated predicted current in real-time. The control performance between dq decoupling control, MPC, and the new algorithm is compared. Besides, the influence of circuit parameters on the robustness of the system is discussed in detail. Finally, to further verify the effectiveness of the proposed method, a hardware-in-the-loop simulation platform is constructed to test the proposed method. The results show that the algorithm can not only suppress LFO but also effectively reduce the harmonic distortion of the current and overcome the shortcomings of more interference caused by the weak decoupling effect of dq axis current of MPC in a traction line-side converter.
- 25 November 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. 7 (3) , 1031-1046
A Model Predictive Control Considering Parameters and System Uncertainties for Suppressing Low-Frequency Oscillations of Traction Dual Rectifiers
Low-frequency oscillation (LFO) overvoltage problems often occur in vehicle-grid coupled systems of high-speed railways. In order to suppress LFO and make up for the problem of weak control effect caused by the large deviation between the model predictive control (MPC) model and the actual system, the uncertainties in parameters and system are enough considered and a combination of MPC and deviation error correction (DEC) is proposed in this paper. First, the error caused by the parameter mismatch, discretization, and sampling process from system noise is analyzed. Then, the DEC is designed to compensate for the calculated predicted current in real-time. The control performance between dq decoupling control, MPC, and the new algorithm is compared. Besides, the influence of circuit parameters on the robustness of the system is discussed in detail. Finally, to further verify the effectiveness of the proposed method, a hardware-in-the-loop simulation platform is constructed to test the proposed method. The results show that the algorithm can not only suppress LFO but also effectively reduce the harmonic distortion of the current and overcome the shortcomings of more interference caused by the weak decoupling effect of dq axis current of MPC in a traction line-side converter.