Sustainable electric vehicles fault detection based on monitoring by deep LearningArchitectures in feature extraction and classification

• Published in: Sustainable energy technologies and assessments Vol. 57
• Main Authors: Wongchai, Anupong, Aoudni, Yassine, Yesubabu, Mannava, Ahmad Reegu, Faheem, Vasantha Gowri, N., Vijayakumar, P.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-06-2023
• Subjects: Data classification; Deep learning; Electric vehicles; Fault detection; Feature extraction; Monitoring; Deep learning; Data classification; Fault detection; Electric vehicles; Feature extraction; Monitoring
• ISSN: 2213-1388
• DOI: 10.1016/j.seta.2023.103178

•This research proposes novel technique in electric vehicle fault detection based on monitoring data classification and feature extraction using deep learning architectures.•Here the input data has been collected as sustainable electric vehicles data using multi-cell parallel electric vehicle and this data has been processed for noise removal and dimensionality reduction.•The processed data features have been extracted using deep stacking auto gradient descent.•Then the extracted features have been classified using monte carlo regressive Gaussian naïve bayes architecture.
Numerous industrial sector paradigms have been altered by the necessity to produce more competitive machinery and the introduction of digital technologies from so-called Industry 4.0. This research proposes novel technique in electric vehicle fault detection based on monitoring data classification and feature extraction using deep learning architectures. The real Proton exchange membrane fuel cell (PEMFC) experiment dataset has been collected as sustainable electric vehicles data using multi-cell parallel electric vehicle and it has been processed for noise removal, dimensionality reduction and extraction using deep stacking auto gradient descent forclassifyingthrough monte Carlo regressive Gaussian naïve bayes architecture.Results of experiments demonstrate that suggested model achieves over 99% accuracy in identifying flooding fault of fuel cell under load-varying situations. The experimental analysis has been carried out in terms of accuracy, robustness, reliability, precision, recall. The proposed technique attained 99% of accuracy, 89% of robustness, 85% of Reliability, 95% of precision and 81% of recall.