Intelligent back-to-back testing with denoising autoencoder-based fault detection and DBSCAN clustering

Back-to-back (B2B) test has been introduced as a pivotal method for ensuring equivalence between model-level and implementation-level behaviour during the validation process of Automotive Software Systems (ASSs). Conventionally, the analysis of B2B execution results depends on the application of exp...

Full description

Saved in:
Bibliographic Details
Main Authors: Mohammad Abboush, Christoph Knieke, Andreas Rausch
Format: Article
Language:English
Published: Elsevier 2025-09-01
Series:Results in Engineering
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2590123025019711
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849424171129372672
author Mohammad Abboush
Christoph Knieke
Andreas Rausch
author_facet Mohammad Abboush
Christoph Knieke
Andreas Rausch
author_sort Mohammad Abboush
collection DOAJ
description Back-to-back (B2B) test has been introduced as a pivotal method for ensuring equivalence between model-level and implementation-level behaviour during the validation process of Automotive Software Systems (ASSs). Conventionally, the analysis of B2B execution results depends on the application of expert knowledge and the utilisation of predetermined thresholds for the identification of failures. This approach, however, is limited in its ability to address the complexities inherent in modern systems, particularly in the presence of non-linear dynamic multivariate behaviour under noisy conditions. To address these limitations, in this study, an intelligent analysis approach is proposed for assisting the test engineers during the analysis process of the B2B test execution results. The approach is capable of automatically detecting and clustering the faults in an efficient manner considering the noisy conditions. To this end, a CNN-LSTM-based denoising autoencoder (DAE) architecture has been developed to accurately detect the faults in the test recordings based on fault-free dataset. Furthermore, an adopting density-based clustering method, i.e., DBSCAN, has been proposed to group the detected faults based on representative features extracted from DAE. The evaluation results demonstrate the superiority of the proposed approach in comparison to state-of-the-art methods in terms of performance and computational cost with F1-score 96.15%, DBI 0.159 and testing time 5.2 ms. Additionally, the experimental findings demonstrate that the proposed approach satisfy the criteria for robustness and generalisation in the presence of noise across diverse driving scenarios with MSE of 0.00312 at 10% noise. Consequently, the proposed approach has the potential to reduce the time and effort associated with traditional analysis while improving the safety and reliability of complex dynamic vehicle systems.
format Article
id doaj-art-34c5052e622c4cb18ed2d1499b7ffeea
institution Kabale University
issn 2590-1230
language English
publishDate 2025-09-01
publisher Elsevier
record_format Article
series Results in Engineering
spelling doaj-art-34c5052e622c4cb18ed2d1499b7ffeea2025-08-20T03:30:19ZengElsevierResults in Engineering2590-12302025-09-012710590010.1016/j.rineng.2025.105900Intelligent back-to-back testing with denoising autoencoder-based fault detection and DBSCAN clusteringMohammad Abboush0Christoph Knieke1Andreas Rausch2Corresponding author.; Technische Universität Clausthal, Institute for Software and Systems Engineering, Arnold-Sommerfeld-Straße 1, Clausthal-Zellerfeld, 38678, Niedersachsen, GermanyTechnische Universität Clausthal, Institute for Software and Systems Engineering, Arnold-Sommerfeld-Straße 1, Clausthal-Zellerfeld, 38678, Niedersachsen, GermanyTechnische Universität Clausthal, Institute for Software and Systems Engineering, Arnold-Sommerfeld-Straße 1, Clausthal-Zellerfeld, 38678, Niedersachsen, GermanyBack-to-back (B2B) test has been introduced as a pivotal method for ensuring equivalence between model-level and implementation-level behaviour during the validation process of Automotive Software Systems (ASSs). Conventionally, the analysis of B2B execution results depends on the application of expert knowledge and the utilisation of predetermined thresholds for the identification of failures. This approach, however, is limited in its ability to address the complexities inherent in modern systems, particularly in the presence of non-linear dynamic multivariate behaviour under noisy conditions. To address these limitations, in this study, an intelligent analysis approach is proposed for assisting the test engineers during the analysis process of the B2B test execution results. The approach is capable of automatically detecting and clustering the faults in an efficient manner considering the noisy conditions. To this end, a CNN-LSTM-based denoising autoencoder (DAE) architecture has been developed to accurately detect the faults in the test recordings based on fault-free dataset. Furthermore, an adopting density-based clustering method, i.e., DBSCAN, has been proposed to group the detected faults based on representative features extracted from DAE. The evaluation results demonstrate the superiority of the proposed approach in comparison to state-of-the-art methods in terms of performance and computational cost with F1-score 96.15%, DBI 0.159 and testing time 5.2 ms. Additionally, the experimental findings demonstrate that the proposed approach satisfy the criteria for robustness and generalisation in the presence of noise across diverse driving scenarios with MSE of 0.00312 at 10% noise. Consequently, the proposed approach has the potential to reduce the time and effort associated with traditional analysis while improving the safety and reliability of complex dynamic vehicle systems.http://www.sciencedirect.com/science/article/pii/S2590123025019711HIL testingFault detection and clusteringAutomotive software systemsReal-time validationBack-to-back testingDeep denoising autoencoder
spellingShingle Mohammad Abboush
Christoph Knieke
Andreas Rausch
Intelligent back-to-back testing with denoising autoencoder-based fault detection and DBSCAN clustering
Results in Engineering
HIL testing
Fault detection and clustering
Automotive software systems
Real-time validation
Back-to-back testing
Deep denoising autoencoder
title Intelligent back-to-back testing with denoising autoencoder-based fault detection and DBSCAN clustering
title_full Intelligent back-to-back testing with denoising autoencoder-based fault detection and DBSCAN clustering
title_fullStr Intelligent back-to-back testing with denoising autoencoder-based fault detection and DBSCAN clustering
title_full_unstemmed Intelligent back-to-back testing with denoising autoencoder-based fault detection and DBSCAN clustering
title_short Intelligent back-to-back testing with denoising autoencoder-based fault detection and DBSCAN clustering
title_sort intelligent back to back testing with denoising autoencoder based fault detection and dbscan clustering
topic HIL testing
Fault detection and clustering
Automotive software systems
Real-time validation
Back-to-back testing
Deep denoising autoencoder
url http://www.sciencedirect.com/science/article/pii/S2590123025019711
work_keys_str_mv AT mohammadabboush intelligentbacktobacktestingwithdenoisingautoencoderbasedfaultdetectionanddbscanclustering
AT christophknieke intelligentbacktobacktestingwithdenoisingautoencoderbasedfaultdetectionanddbscanclustering
AT andreasrausch intelligentbacktobacktestingwithdenoisingautoencoderbasedfaultdetectionanddbscanclustering