Wissenschaftliche Reihe Fahrzeugtechnik Universität Stuttgart Tunan Shen Diagnosis of the Powertrain Systems for Autonomous Electric Vehicles Wissenschaftliche Reihe Fahrzeugtechnik Universität Stuttgart Reihe herausgegeben von Michael Bargende, Stuttgart, Deutschland Hans-Christian Reuss, Stuttgart, Deutschland Jochen Wiedemann, Stuttgart, Deutschland Das Institut für Fahrzeugtechnik Stuttgart (IFS) an der Universität Stuttgart erforscht, entwickelt, appliziert und erprobt, in enger Zusammenarbeit mit der Industrie, Elemente bzw. Technologien aus dem Bereich moderner Fahrzeugkonzepte. Das Institut gliedert sich in die drei Bereiche Kraftfahrwesen, Fahrzeugantriebe und Kraftfahrzeug-Mechatronik. Aufgabe dieser Bereiche ist die Ausarbeitung des Themengebietes im Prüfstandsbetrieb, in Theorie und Simu- lation.
Schwerpunkte des Kraftfahrwesens sind hierbei die Aerodynamik, Akustik (NVH), Fahrdynamik und Fahrermodellierung, Leichtbau, Sicherheit, Kraftüber- tragung sowie Energie und Thermomanagement – auch in Verbindung mit hybri- den und batterieelektrischen Fahrzeugkonzepten. Der Bereich Fahrzeugantriebe widmet sich den Themen Brennverfahrensentwicklung einschließlich Regelungs- und Steuerungskonzeptionen bei zugleich minimierten Emissionen, komplexe Abgasnachbehandlung, Aufladesysteme und -strategien, Hybridsysteme und Betriebsstrategien sowie mechanisch-akustischen Fragestellungen. Themen der Kraftfahrzeug-Mechatronik sind die Antriebsstrangregelung/Hybride, Elektromo- bilität, Bordnetz und Energiemanagement, Funktions- und Softwareentwick- lung sowie Test und Diagnose. Die Erfüllung dieser Aufgaben wird prüf- standsseitig neben vielem anderen unterstützt durch 19 Motorenprüfstände, zwei Rollenprüfstände, einen 1:1-Fahrsimulator, einen Antriebsstrangprüfs- tand, einen Thermowindkanal sowie einen 1:1-Aeroakustikwindkanal.
Die wis- senschaftliche Reihe „Fahrzeugtechnik Universität Stuttgart“ präsentiert über die am Institut entstandenen Promotionen die hervorragenden Arbeitsergebnisse der Forschungstätigkeiten am IFS. Reihe herausgegeben von Prof. Michael Bargende Prof. Hans-Christian Reuss Lehrstuhl Fahrzeugantriebe Lehrstuhl Kraftfahrzeugmechatronik Institut für Fahrzeugtechnik Stuttgart Institut für Fahrzeugtechnik Stuttgart Universität Stuttgart Universität Stuttgart Stuttgart, Deutschland Stuttgart, Deutschland Prof.
Jochen Wiedemann Lehrstuhl Kraftfahrwesen Institut für Fahrzeugtechnik Stuttgart Universität Stuttgart Stuttgart, Deutschland Weitere Bände in der Reihe https://link.com/bookseries/13535 Tunan Shen Diagnosis of the Powertrain Systems for Autonomous Electric Vehicles Tunan Shen Institute of Automotive Engineering (IFS), Chair of Automotive Mechatronics University of Stuttgart Stuttgart, Germany Zugl.: Dissertation Universität Stuttgart, 2021 D93 ISSN 2567-0042 ISSN 2567-0352 (electronic) Wissenschaftliche Reihe Fahrzeugtechnik Universität Stuttgart ISBN 978-3-658-36991-0 ISBN 978-3-658-36992-7 (eBook) https://doi.1007/978-3-658-36992-7 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2022 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprint- ing, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Responsible Editor: Stefanie Eggert This Springer Vieweg imprint is published by the registered company Springer Fachmedien Wiesbaden GmbH part of Springer Nature.
The registered company address is: Abraham-Lincoln-Str. 46, 65189 Wiesbaden, Germany Preface I believe that the electric vehicle revolution is coming and automated driving will come soon. Then, most people may not own a car in future any more, but hire self-driving, electric ride-shares to get around. For those autonomous electric vehicles, which owned by ride-share companies, the reliability and availability are more important than today’s private owned cars.
I want to be a part of the change, so I decided to pursue my PhD in the area of diagnosis of powertrain system for autonomous electric vehicles since 2017, after three years working at Robert Bosch GmbH. My deep gratitude goes to my supervisor Prof. Hans-Christian Reussfor his endless support, enthusiasm, knowledge and friendship. I am extremely grateful to my supervisors at Robert Bosch GmbH Dr.
Ahmet Kilic and Dr. Christian Thulfaut for the valuable support and motivation. Their insights and advice give me a great help. Special thanks to Dr.
Norbert Müller and Dr. Achim Henkel for giving me the chance to carry out this work at Corporate Sector Research and Advance Engi- neering at Robert Bosch GmbH in Renningen. My appreciation also extends to Adam Babik, Dr. Andreas Vogt, Dr.
Andreas Schönknecht, Daniel Acs, Dr. Deng Shi, Erik Hoevenaars, Dr. Paul Mehringer, Rajaram Suresh and Yuping Chen for your valuable comments on this work and all my colleagues for the wonderful cooperation as well. Furthermore, I want to thank my wife Shiang and my parents.
You believed in me when I was doubt. You kept me motivated through my darkest thoughts. Finally, I would like to thank my daughter Cindy. Thank you for coming into my life.
Thank you for making me smile like crazy. Thank you for making me happy. Leonberg Tunan Shen Inhaltsverzeichnis Preface.3 Contributions of the Thesis .4 Organization of the Thesis .5 Scope of the Thesis. 10 3 Background and State of the Art .1 Fault Diagnostic Methods .1 Signal-based Fault Detection Methods .2 Model-based Fault Detection Methods .3 Data-based Fault Detection Methods.2 Signal Processing Techniques .1 Time Domain Features.3 Frequency Domain Features.5 Time Frequency Domain Features .3 Machine Learning Algorithms .3 Artificial Neural Network.
23 4 Diagnosis of Electrical Faults in Electric Machines .1 Related Works and Current Challenges .2 Contributions of the Thesis .3 Analytical Modeling of Faults.1 Analytical Modeling of a Healthy PMSM .2 Analytical Modeling of PMSM with TSC .3 Analytical Modeling of PMSM with PSC .4 Analytical Modeling of PMSM with UWR .5 Analytical Modeling of PMSM with Sensor Faults .4 Analysis of the Behavior of a PMSM in Different Conditions .7 Physical Model based Diagnostic Model .8 Self Condition Monitoring (SCM) Diagnostic Model .9 Fleet Data-based Fault Diagnostic Model .10 Multi-stage Diagnostic Concept. 65 5 Diagnosis of Mechanical Faults in Electric Machines.1 Fault Mechanisms of Bearing .3 Current Challenges and Contribution of the Thesis .2 Contributions of the Thesis .4 Data Set Description .3 Evaluation of Features .6 Validation with Other eAxles .101 6 Conclusion and Outlook.1 Failure mechanism of a battery fire .2 Distribution of fragile components.3 Distribution of failed components in electric machines .1 Scheme of a signal-based fault detection method .2 Scheme of a model-based fault detection method .3 Scheme of a data-based fault detection method .4 Envelope of a vibration signal.5 An example of STFT. (a) Quadratic chirp signal; (b) STFT of quadratic chirp signal.6 A schematic structure of decision tree .7 (a) Distance-based anomaly detection; (b) Density-based anomaly detection.8 Diagram of an artificial neuron .9 Structure of neural network .1 Equivalent circuit of healthy and faulty stators of PMSM.2 The difference of a current sensor in healthy and faulty conditions.3 Operating area of PMSM.4 Current and torque of machine in different conditions at operating point Tre f = 100 Nm, nmech = 5000 rpm.5 Comparison of the three phase currents of healthy and faulty con- ditions at operating point Tre f = 40 Nm, nmech = 4000 rpm .6 Comparison of symptoms in different conditions .7 The tree structure of decision tree model .8 Normalized confusion matrices of the physical model on (a) trai- ning and (b) test data .9 Wrong predictions of the physical model .10 Distribution of distance .11 Confusion matrices of the SCM model on (a) training and validation data and (b) test data.12 Wrong predictions of the SCM model .13 Confusion matrices of NN model on (a) training, (b) validation and (c) test data.14 Comparison of the probabilities of right and wrong predictions .15 Comparison the probabilities of right and wrong predictions .16 Confusion matrices of (a) physical model, (b) SCM model and (c)NN model on test data .17 Confusion matrices of the physical model at (a) frequently driven point, (b) low speed point, (c) high speed point.18 Confusion matrices of combined models.19 Multi-class confusion matrices of combined model at (a) 1st stage, (b) 2nd stage, and (c) 3rd stage.1 Rolling element bearing. (a) Structure of a rolling element bearing; (b) Pictures of defective bearings.2 Ramp tests over lifetime.3 Rms at different measurement time and speeds.4 The noisy data.5 The filtered data.6 Rms of bearing 2-3 in XJTU-SY data set.7 Rms of an eAxle.8 Process from raw data to health indicator .9 Normalized Rms in different clusters.10 Rms as a health indicator.11 Cage characteristic feature Pwr f t f of bearing 2-3.12 Cage characteristic feature Env f t f of bearing 2-3.13 Spectrogram of a ramp test.14 The first three harmonics of FTF.15 Cage characteristic feature Pwr f t f of an eAxle.16 Cage characteristic feature Env f t f of an eAxle.17 Spectrogram of ramp test No.18 Spectrogram of ramp test No.19 Spectrogram of ramp test No.20 Medianspec over the time.21 Comparison of features.22 The ranking of sensitivity.23 The ranking of monotonicity.24 Feature P2P of four eAxles.25 Feature Env f t f of four eAxles.26 Feature Medianspec of four eAxles.27 Anomaly score of eAxle1.28 Prediction of the Medianspec curves.29 Comparison of the synthetic and true Medianspec curves.30 Prediction of the synthetic Medianspec curve.1 Time domain features .3 Characteristic frequencies of bearing faults.4 Typical activation functions.1 Parameters of PMSM.2 Extracted time domain and statistical features.3 Comparison of time domain and statistical features in different conditions .4 Comparison of harmonic features in different conditions .5 The number of samples .7 Hyperparameters of NN model .8 Evaluation of single models .9 Evaluation of combined models.10 Evaluation of models with unknown faults.1 A short overview of open accessed bearing data sets .2 A list of features used for bearing RUL prediction.3 Comparison of two data sets.4 A list of extracted features from the eAxle data set.