Tharun Kommaddi hat seine Masterarbeit an der TU Clausthal erfolgreich abgeschlossen. Das Thema seiner Arbeit lautet: “Detecting and Clustering Unknown Concurrent faults of Automotive Software Systems under Noisy Conditions based on Hardware-in-the-Loop Tests and Deep Learning Technique“.
The significance of this research has been highlighted by the growing complexity of Automotive Software Systems (ASSs) and the resulting requirement for efficient fault detection and clustering solutions. It is not a long-term solution to rely on human skill to understand enormous volumes of multivariate time series data during Hardware-in-the-Loop (HIL) simulations. Additionally, the use of noisy and unlabelled data makes it more difficult to use current automatic defect diagnostic techniques. By creating a novel deep-learning model for automatically identifying and clustering unknown faults in ASSs, this study aims to overcome these difficulties. The proposed model uses Gated Recurrent Unit-based Denoising Autoencoder (GRU-DAE) for representative features extraction and K-Means method to detect and cluster the faults. The research demonstrates the applicability and dependability of the suggested model by the analysis of digital test drive data and real-time Hardware-in-the-Loop (HIL) simulation performed in a variety of settings. Finally, by offering a more effective technique for handling the complexity of ASSs and lowering the likelihood of failures, the suggested method contributes to improving the safety and dependability of ASSs. This study emphasizes the potential of deep learning methods as vital tools for fault detection and clustering in the automobile industry.