EPE Association: EPE 2023 - LS7e: Data Analysis, Artificial Intelligence and Communication

EPE 2023 - LS7e: Data Analysis, Artificial Intelligence and Communication
You are here: EPE Documents > 01 - EPE & EPE ECCE Conference Proceedings > EPE 2023 ECCE Europe - Conference > EPE 2023 - Topic 10: Data Analysis, Artificial Intelligence and Communication > EPE 2023 - LS7e: Data Analysis, Artificial Intelligence and Communication

   [return to parent folder]

   Comparison Study on Parametric Fault Diagnosis Using BPNN, SVM and SDAE for DC-DC Converters in Aircraft
By Ting WANG, Jiacheng SUN, Wenli YAO, Xiaobin ZHANG, Weilin LI, Yufeng WANG [View]
[Download]

Abstract: Effective fault diagnosis for mission-critical and safety-critical systems, such as aircraft electric powersystem, has been an essential and mandatory technique to reduce failure rate and prevent unscheduled shutdown. This paper aims to compare the performance of three efficient fault classifiers, BPNN, SVM and SDAE, in parametric fault diagnosis for the boost DC-DC converter in aircraft. The training set and test set are collected based on the fitting of NASA datasets of electrolytic capacitors/MOSFET and a boost DC-DC converter simulation system. Effective fault features are extracted from four node signals using time-domain and statistical analysis. Seven kinds of faults of electrolytic capacitor and power MOSFET were studied. The simulation results show that SVM and SDAE have a higher classification accuracy for parametric faults, such as the component degradation of electrolytic capacitor and power MOSFET, but BPNN has fast diagnosis, more suitable for cases with small data volume.

   Health Monitoring Framework for Electric Vehicle Drive Train in Digital Twin
By Varaha Satya Bharath KURUKURU, Mohammed Ali KHAN, Rupam SINGH [View]
[Download]

Abstract: As electric vehicles (EVs) continue to evolve and become more intricate, it becomes increasingly important to monitor their health continuously to ensure both safe operation and optimal performance. To address this need, this research paper proposes a comprehensive health monitoring framework that leverages the concept of Digital Twin (DT). The DT incorporates a bond graph (BG) model, which accurately represents the intricate structure and functionality of the EV drivetrain. Additionally, the framework utilizes Support Vector Data Description (SVDD) to train and classify measured data effectively, enabling efficient fault detection and diagnosis. By integrating the developed BG model and SVDD into the digital twin, the framework enables real-time monitoring and predictive analysis of the EV's health status. The simulation results demonstrate the effectiveness of this framework, showcasing high accuracies of 98.7\% during training and 96.21\% during testing. These results validate the potential of the proposed approach to ensure the reliable and efficient operation of EVs while also minimizing the risk of malfunctions and ensuring a safe driving experience for users.

   Ultrafast Feature Extraction for Lithium-Ion Battery Health Assessment
By Xin SUI, Shan HE, Remus TEODORESCU [View]
[Download]

Abstract: Machine learning (ML) becomes an important technology in battery health assessment. The mapping from feature usually extracted from charging voltage or temperature to unmeasurable state of health (SOH) can be found by training a ML-based SOH estimator. However, the feature may become invalid when operation conditions change or be inaccessible from incomplete charging. For tackling these challenges, various entropies are investigated thoughtfully. Afterwards, spectral entropy and its variants, i.e., composite multi-scale entropy and hierarchical entropy are screened out. Ultrafast SOH feature extraction is therefore achieved where only 2 seconds of voltage data is needed. Finally, the effectiveness of the proposed method is verified by using the accelerated aging dataset from NMC batteries.

EPE 2023 - LS7e: Data Analysis, Artificial Intelligence and Communication
You are here: EPE Documents > 01 - EPE & EPE ECCE Conference Proceedings > EPE 2023 ECCE Europe - Conference > EPE 2023 - Topic 10: Data Analysis, Artificial Intelligence and Communication > EPE 2023 - LS7e: Data Analysis, Artificial Intelligence and Communication
	[return to parent folder]

	Comparison Study on Parametric Fault Diagnosis Using BPNN, SVM and SDAE for DC-DC Converters in Aircraft By Ting WANG, Jiacheng SUN, Wenli YAO, Xiaobin ZHANG, Weilin LI, Yufeng WANG	[View] [Download]
Abstract: Effective fault diagnosis for mission-critical and safety-critical systems, such as aircraft electric powersystem, has been an essential and mandatory technique to reduce failure rate and prevent unscheduled shutdown. This paper aims to compare the performance of three efficient fault classifiers, BPNN, SVM and SDAE, in parametric fault diagnosis for the boost DC-DC converter in aircraft. The training set and test set are collected based on the fitting of NASA datasets of electrolytic capacitors/MOSFET and a boost DC-DC converter simulation system. Effective fault features are extracted from four node signals using time-domain and statistical analysis. Seven kinds of faults of electrolytic capacitor and power MOSFET were studied. The simulation results show that SVM and SDAE have a higher classification accuracy for parametric faults, such as the component degradation of electrolytic capacitor and power MOSFET, but BPNN has fast diagnosis, more suitable for cases with small data volume.

	Health Monitoring Framework for Electric Vehicle Drive Train in Digital Twin By Varaha Satya Bharath KURUKURU, Mohammed Ali KHAN, Rupam SINGH	[View] [Download]
Abstract: As electric vehicles (EVs) continue to evolve and become more intricate, it becomes increasingly important to monitor their health continuously to ensure both safe operation and optimal performance. To address this need, this research paper proposes a comprehensive health monitoring framework that leverages the concept of Digital Twin (DT). The DT incorporates a bond graph (BG) model, which accurately represents the intricate structure and functionality of the EV drivetrain. Additionally, the framework utilizes Support Vector Data Description (SVDD) to train and classify measured data effectively, enabling efficient fault detection and diagnosis. By integrating the developed BG model and SVDD into the digital twin, the framework enables real-time monitoring and predictive analysis of the EV's health status. The simulation results demonstrate the effectiveness of this framework, showcasing high accuracies of 98.7\% during training and 96.21\% during testing. These results validate the potential of the proposed approach to ensure the reliable and efficient operation of EVs while also minimizing the risk of malfunctions and ensuring a safe driving experience for users.

	Ultrafast Feature Extraction for Lithium-Ion Battery Health Assessment By Xin SUI, Shan HE, Remus TEODORESCU	[View] [Download]
Abstract: Machine learning (ML) becomes an important technology in battery health assessment. The mapping from feature usually extracted from charging voltage or temperature to unmeasurable state of health (SOH) can be found by training a ML-based SOH estimator. However, the feature may become invalid when operation conditions change or be inaccessible from incomplete charging. For tackling these challenges, various entropies are investigated thoughtfully. Afterwards, spectral entropy and its variants, i.e., composite multi-scale entropy and hierarchical entropy are screened out. Ultrafast SOH feature extraction is therefore achieved where only 2 seconds of voltage data is needed. Finally, the effectiveness of the proposed method is verified by using the accelerated aging dataset from NMC batteries.