{"id":{"repo_id":"tu-berlin","oai_identifier":"oai:depositonce.tu-berlin.de:11303/25852"},"canonical_url":"https://search.dev.ndltd.org/etd/tu-berlin/oai:depositonce.tu-berlin.de:11303/25852","repository":{"repo_id":"tu-berlin","name":"Technische Universität Berlin","base_url":"https://api-depositonce.tu-berlin.de/server/oai/request"},"display":{"title":"Autoencoder-based multivariate time series anomaly detection and clustering for diagnosis of mechatronic systems","abstract":"In this thesis, the ability of a specific type of Deep Artificial Neural Network (DANN) called the Autoencoder to quantify the degree of fault, failure or malfunction (anomaly) of a mechatronic system using raw sensory data is analyzed. The Autoencoder approach follows the one-class classification paradigm and is trained unsupervised by reconstructing its own input using fault-free data only. For quantification of the faults or deterioration of the system, a health index is computed from an anomaly score, which is usually based on the reconstruction error. Four different distance metrics, Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mahalanobis distance and Cosine similarity are evaluated for the error calculation. Additionally, the error calculation is performed and evaluated using the latent variables, also called representation, of the Autoencoder. Time windows with different lengths from multivariate time-series data as its input are used during training and inferencing of the Artificial Neural Network (ANN) for analyzing the fault detection performances’ time dependence with transient data. The fault detection performance is evaluated using the Area Under Curve Receiver Operating Characteristic (AUCROC) on the three publicly available datasets New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS), Tennessee Eastman Process (TEP) and Gas Sensor Array Drift (GSAD) and compared against the Principal Component Analysis (PCA) as the model of the normal system instead of the Autoencoder. Various Autoencoder regularization terms and pre-layer e.g. Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) in combination with different activation function, objective function, batch size and other hyperparameter are varied on all datasets and health index computation methods in an extensive experimental series. With the best-performing combinations of Autoencoder types and hyperparameter the fault detection performance is further optimized by applying automated clustering on the training data. The clustering is done by the self-developed Autoencoder Based Iterative Modeling and multivariate time-series Clustering Algorithm (ABIMCA) approach, which shows a strong increase in the detection quality and can be used as a fault detection approach itself.","abstract_html":"In this thesis, the ability of a specific type of Deep Artificial Neural Network (DANN) called the Autoencoder to quantify the degree of fault, failure or malfunction (anomaly) of a mechatronic system using raw sensory data is analyzed. The Autoencoder approach follows the one-class classification paradigm and is trained unsupervised by reconstructing its own input using fault-free data only. For quantification of the faults or deterioration of the system, a health index is computed from an anomaly score, which is usually based on the reconstruction error. Four different distance metrics, Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mahalanobis distance and Cosine similarity are evaluated for the error calculation. Additionally, the error calculation is performed and evaluated using the latent variables, also called representation, of the Autoencoder. Time windows with different lengths from multivariate time-series data as its input are used during training and inferencing of the Artificial Neural Network (ANN) for analyzing the fault detection performances’ time dependence with transient data. The fault detection performance is evaluated using the Area Under Curve Receiver Operating Characteristic (AUCROC) on the three publicly available datasets New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS), Tennessee Eastman Process (TEP) and Gas Sensor Array Drift (GSAD) and compared against the Principal Component Analysis (PCA) as the model of the normal system instead of the Autoencoder. Various Autoencoder regularization terms and pre-layer e.g. Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) in combination with different activation function, objective function, batch size and other hyperparameter are varied on all datasets and health index computation methods in an extensive experimental series. With the best-performing combinations of Autoencoder types and hyperparameter the fault detection performance is further optimized by applying automated clustering on the training data. The clustering is done by the self-developed Autoencoder Based Iterative Modeling and multivariate time-series Clustering Algorithm (ABIMCA) approach, which shows a strong increase in the detection quality and can be used as a fault detection approach itself.","abstract_has_math":false,"creators":["Köhne, Jonas"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Gühmann, Clemens"],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-07-27T21:28:31Z","subjects":[],"languages":["en"],"rights":[],"rights_urls":["https://creativecommons.org/licenses/by/4.0/"],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.14279/depositonce-24677"],"render_values":[{"text":"https://doi.org/10.14279/depositonce-24677","href":"https://doi.org/10.14279/depositonce-24677","code":true}]}]},"links":{"outbound_url":"https://depositonce.tu-berlin.de/handle/11303/25852","outbound_label":"Repository record","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Gühmann, Clemens"]},{"key":"dc:creator","label":"Author","values":["Köhne, Jonas"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-12-19T15:16:36Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2025-12-19T15:16:36Z"]},{"key":"dc:date.issued","label":"Date","values":["2025"]},{"key":"dc:type","label":"Dc Type","values":["Doctoral Thesis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]},{"key":"dc:rights.uri","label":"Rights URI","values":["https://creativecommons.org/licenses/by/4.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://depositonce.tu-berlin.de/handle/11303/25852","https://doi.org/10.14279/depositonce-24677"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["In this thesis, the ability of a specific type of Deep Artificial Neural Network (DANN) called the Autoencoder to quantify the degree of fault, failure or malfunction (anomaly) of a mechatronic system using raw sensory data is analyzed. The Autoencoder approach follows the one-class classification paradigm and is trained unsupervised by reconstructing its own input using fault-free data only. For quantification of the faults or deterioration of the system, a health index is computed from an anomaly score, which is usually based on the reconstruction error. Four different distance metrics, Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mahalanobis distance and Cosine similarity are evaluated for the error calculation. Additionally, the error calculation is performed and evaluated using the latent variables, also called representation, of the Autoencoder. Time windows with different lengths from multivariate time-series data as its input are used during training and inferencing of the Artificial Neural Network (ANN) for analyzing the fault detection performances’ time dependence with transient data. The fault detection performance is evaluated using the Area Under Curve Receiver Operating Characteristic (AUCROC) on the three publicly available datasets New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS), Tennessee Eastman Process (TEP) and Gas Sensor Array Drift (GSAD) and compared against the Principal Component Analysis (PCA) as the model of the normal system instead of the Autoencoder. Various Autoencoder regularization terms and pre-layer e.g. Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) in combination with different activation function, objective function, batch size and other hyperparameter are varied on all datasets and health index computation methods in an extensive experimental series. With the best-performing combinations of Autoencoder types and hyperparameter the fault detection performance is further optimized by applying automated clustering on the training data. The clustering is done by the self-developed Autoencoder Based Iterative Modeling and multivariate time-series Clustering Algorithm (ABIMCA) approach, which shows a strong increase in the detection quality and can be used as a fault detection approach itself.","In dieser Arbeit wird die Fähigkeit eines speziellen Typs eines Deep Artificial Neural Network (DANN) analysiert, des sogenannten Autoencoders, den Grad eines Fehlers, Ausfalls oder einer Fehlfunktion eines mechatronischen Systems anhand von sensorischen Rohdaten zu quantifizieren. Der Autoencoder-Ansatz folgt dem Ein-Klassen-Klassifizierungsparadigma und wird unüberwacht trainiert, indem er seine Eingangsdaten zu rekonstruieren versucht, welche nur vom fehlerfreier Zustand des Systems stammen. Zur Quantifizierung der Abweichung des Systems vom Normal-zustand wird ein Gesundheitsindex aus einer Anomaliebewertung berechnet, welche in der Regel auf dem Rekonstruktionsfehler basiert. Für die Fehlerberechnung werden vier verschiedene Distanzmetriken ausgewertet: Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mahalanobis-Distanz und Kosinus-Ähnlichkeit. Die Fehlerberechnung wird auch unter Verwendung der latenten Variablen des Autoenco-ders, auch Repräsentation genannt, durchgeführt und ausgewertet. Zeitfenster mit unterschiedlichen Längen aus multivariaten Zeitreihendaten werden während des Trainings und der Inferenz des künstlichen neuronalen Netzes (ANN) verwendet, um die Zeitabhängigkeit der Fehlererkennungsleistung mit transienten Daten zu analysieren. Die Fehlererkennungsleistung wird mit Hilfe der Area Under Curve Receiver Operating Characteristic (AUCROC) an den drei öffentlich verfügbaren Datensätzen New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS), Tennessee Eastman Process (TEP) und Gas Sensor Array Drift (GSAD) bewertet und mit der Principal Component Analysis (PCA) als Modell des normalen Systems anstelle des Autoencoders verglichen. In einer umfangreichen Versuchsreihe werden verschiedene Autoencoder-Regularisierungsbedingungen und vorgelagerte Artificial Neural Network (ANN)-Schichten, z.B. Recurrent Neural Network (RNN) und Convolutional Neural Network (CNN), in Kombination mit verschiedenen Aktivierungsfunktionen, Zielfunktionen, Batch-Größen und anderen Hyperparametern auf allen Datensätzen und Gesundheitsindex-Berechnungsmethoden variiert. Mit den besten Kombinationen von Autoencoder-Typen und Hyperparametern wird durch automatisches Clustern der Trainingsdaten die Fehlererkennung weiter optimiert. Das Clustering erfolgt durch den selbst entwickelten Autoencoder Based Iterative Modeling and multivariate time-series Clustering Algorithm (ABIMCA) Ansatz, der eine starke Steigerung der Fehlererkennungsleistung zeigt und selbst als Fehlererkennungsansatz verwendet werden kann."]},{"key":"dc:title","label":"Title","values":["Autoencoder-based multivariate time series anomaly detection and clustering for diagnosis of mechatronic systems"]}]}],"canonical_facts":{"dc:contributor.advisor":["Gühmann, Clemens"],"dc:creator":["Köhne, Jonas"],"dc:date.accessioned":["2025-12-19T15:16:36Z"],"dc:date.available":["2025-12-19T15:16:36Z"],"dc:date.issued":["2025"],"dc:description.abstract":["In this thesis, the ability of a specific type of Deep Artificial Neural Network (DANN) called the Autoencoder to quantify the degree of fault, failure or malfunction (anomaly) of a mechatronic system using raw sensory data is analyzed. The Autoencoder approach follows the one-class classification paradigm and is trained unsupervised by reconstructing its own input using fault-free data only. For quantification of the faults or deterioration of the system, a health index is computed from an anomaly score, which is usually based on the reconstruction error. Four different distance metrics, Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mahalanobis distance and Cosine similarity are evaluated for the error calculation. Additionally, the error calculation is performed and evaluated using the latent variables, also called representation, of the Autoencoder. Time windows with different lengths from multivariate time-series data as its input are used during training and inferencing of the Artificial Neural Network (ANN) for analyzing the fault detection performances’ time dependence with transient data. The fault detection performance is evaluated using the Area Under Curve Receiver Operating Characteristic (AUCROC) on the three publicly available datasets New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS), Tennessee Eastman Process (TEP) and Gas Sensor Array Drift (GSAD) and compared against the Principal Component Analysis (PCA) as the model of the normal system instead of the Autoencoder. Various Autoencoder regularization terms and pre-layer e.g. Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) in combination with different activation function, objective function, batch size and other hyperparameter are varied on all datasets and health index computation methods in an extensive experimental series. With the best-performing combinations of Autoencoder types and hyperparameter the fault detection performance is further optimized by applying automated clustering on the training data. The clustering is done by the self-developed Autoencoder Based Iterative Modeling and multivariate time-series Clustering Algorithm (ABIMCA) approach, which shows a strong increase in the detection quality and can be used as a fault detection approach itself.","In dieser Arbeit wird die Fähigkeit eines speziellen Typs eines Deep Artificial Neural Network (DANN) analysiert, des sogenannten Autoencoders, den Grad eines Fehlers, Ausfalls oder einer Fehlfunktion eines mechatronischen Systems anhand von sensorischen Rohdaten zu quantifizieren. Der Autoencoder-Ansatz folgt dem Ein-Klassen-Klassifizierungsparadigma und wird unüberwacht trainiert, indem er seine Eingangsdaten zu rekonstruieren versucht, welche nur vom fehlerfreier Zustand des Systems stammen. Zur Quantifizierung der Abweichung des Systems vom Normal-zustand wird ein Gesundheitsindex aus einer Anomaliebewertung berechnet, welche in der Regel auf dem Rekonstruktionsfehler basiert. Für die Fehlerberechnung werden vier verschiedene Distanzmetriken ausgewertet: Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mahalanobis-Distanz und Kosinus-Ähnlichkeit. Die Fehlerberechnung wird auch unter Verwendung der latenten Variablen des Autoenco-ders, auch Repräsentation genannt, durchgeführt und ausgewertet. Zeitfenster mit unterschiedlichen Längen aus multivariaten Zeitreihendaten werden während des Trainings und der Inferenz des künstlichen neuronalen Netzes (ANN) verwendet, um die Zeitabhängigkeit der Fehlererkennungsleistung mit transienten Daten zu analysieren. Die Fehlererkennungsleistung wird mit Hilfe der Area Under Curve Receiver Operating Characteristic (AUCROC) an den drei öffentlich verfügbaren Datensätzen New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS), Tennessee Eastman Process (TEP) und Gas Sensor Array Drift (GSAD) bewertet und mit der Principal Component Analysis (PCA) als Modell des normalen Systems anstelle des Autoencoders verglichen. In einer umfangreichen Versuchsreihe werden verschiedene Autoencoder-Regularisierungsbedingungen und vorgelagerte Artificial Neural Network (ANN)-Schichten, z.B. Recurrent Neural Network (RNN) und Convolutional Neural Network (CNN), in Kombination mit verschiedenen Aktivierungsfunktionen, Zielfunktionen, Batch-Größen und anderen Hyperparametern auf allen Datensätzen und Gesundheitsindex-Berechnungsmethoden variiert. Mit den besten Kombinationen von Autoencoder-Typen und Hyperparametern wird durch automatisches Clustern der Trainingsdaten die Fehlererkennung weiter optimiert. Das Clustering erfolgt durch den selbst entwickelten Autoencoder Based Iterative Modeling and multivariate time-series Clustering Algorithm (ABIMCA) Ansatz, der eine starke Steigerung der Fehlererkennungsleistung zeigt und selbst als Fehlererkennungsansatz verwendet werden kann."],"dc:identifier.uri":["https://depositonce.tu-berlin.de/handle/11303/25852","https://doi.org/10.14279/depositonce-24677"],"dc:language.iso":["en"],"dc:rights.uri":["https://creativecommons.org/licenses/by/4.0/"],"dc:title":["Autoencoder-based multivariate time series anomaly detection and clustering for diagnosis of mechatronic systems"],"dc:type":["Doctoral Thesis"]},"updated_at":"2026-07-27T21:28:31Z"}