Health Management, Fault Diagnosis and Supervisory Control for Intelligent Engineering Systems

Health Management, Fault Diagnosis and Supervisory Control for Intelligent Engineering Systems. Summer School Course of 30 hours for the Nanjing University of Aeronautics and Astronautics - NUAA.
 
 

Prerequisites

Participants are expected to have basic knowledge of automatic control systems, as well as fundamental proficiency in MATLAB and Simulink software environments. Familiarity with linear algebra, system dynamics, and elementary signal processing concepts is also recommended to benefit from the course content fully. Participants are recommended to complete the free online tutorials MATLAB Onramp (https://matlabacademy.mathworks.com/details/matlab-onramp/gettingstarted/) and Simulink Onramp (https://matlabacademy.mathworks.com/details/simulink-onramp/simulink) provided by MathWorks before the course.

Course Abstract

The NUAA Summer School 2026 is an advanced course specifically tailored for students in Aerospace Engineering, Automatic Control, and related fields. The course provides comprehensive coverage of robust and fault-tolerant control methodologies, combining theoretical foundations with practical applications relevant to aerospace systems. Participants will explore key concepts, including model-based fault detection and isolation, robust controller design, and fault-tolerant control strategies, supported by extensive MATLAB and Simulink examples. The integration of model-based and data-driven methods represents a cutting-edge approach that can significantly enhance the reliability and performance of fault detection systems, which is particularly critical in aerospace engineering applications where safety and precision are paramount. Recent advancements in artificial intelligence and machine learning have enabled unprecedented accuracy in fault detection and diagnosis, particularly through deep learning and advanced classification algorithms. Participants will gain insight into these contemporary developments and their practical implementation. Real-world case studies and practical exercises will illustrate how to apply these techniques to critical aerospace scenarios, enhancing students' analytical and problem-solving skills. By attending the course, students will be equipped with essential tools and advanced knowledge necessary to design, simulate, and validate sophisticated control systems, significantly contributing to their academic and professional development in international aerospace engineering contexts. This summer school also represents an excellent networking opportunity, enabling participants to build lasting academic and professional connections within the international aerospace control community. A final assignment will allow participants to complete a self-evaluation of the topics and skills acquired.

Registration Information

For registration details, deadlines, and any further inquiries, please get in touch with the administration office of the Nanjing University of Aeronautics and Astronautics (NUAA).

Learning Objectives and Expected Outcomes

The summer school aims to provide participants with advanced theoretical knowledge and practical skills concerning robust and fault-tolerant control methodologies, with specific emphasis on aerospace applications.

Upon successful completion of the course, participants will be able to:

• Understand the theoretical principles behind robust and fault-tolerant control systems and their role in aerospace engineering applications.
• Design and implement robust control strategies capable of handling uncertainties and disturbances typically encountered in aerospace contexts.
• Apply advanced fault diagnosis and fault-tolerant control techniques to ensure the reliability, safety, and performance of aerospace systems.
• Analyse and interpret simulation results, identifying critical performance metrics and system limitations.
• Demonstrate proficiency in MATLAB and Simulink environments, effectively developing and evaluating control system solutions for real-world aerospace problems.
• Collaborate effectively within international research teams, enhancing communication skills and intercultural competencies crucial for future professional careers in global aerospace engineering projects.

Participants will receive structured feedback throughout the course, supporting the continuous assessment and development of their control engineering knowledge and skills. At the end of the course, learning outcomes will be assessed through a final test consisting of 30 multiple-choice questions covering the main topics addressed.

For self-assessment and preparation for the final examination, students may download a sample set of 10 multiple-choice questions (PDF file link), together with a separate file containing the correct answers (PDF file link).

Lecture General Programme and Schedule

Course duration: 10 lectures, approximately 2.5 hours each (total scheduled time: about 32 hours)
Daily schedule: Monday-Friday, 8:00 am - 11:00 am (including a short break). The indicated time refers to Central European Time (CET)/Central European Summer Time (CEST), corresponding to Rome local time. In Nanjing (China Standard Time, CST, UTC+8), this is 2:00 pm - 5:00 pm.

Week 1

• Lecture 1 (Monday): Introduction to Fault Detection in Condition Monitoring
  • Overview of fault detection techniques
  • Condition monitoring principles
  • Fault detection applications in aerospace
  • Introduction to MATLAB/Simulink environment
• Lecture 2 (Tuesday): Model-Based Fault Detection Methods (I)
  • Analytical redundancy
  • Residual generation methods
  • Observer-based approaches (Part 1)
  • MATLAB practical examples
• Lecture 3 (Wednesday): Model-Based Fault Detection Methods (II)
  • Observer-based approaches (Part 2)
  • Parity-space methods
  • MATLAB and Simulink exercises
• Lecture 4 (Thursday): Robust Fault Detection Methods
  • Robustness concepts and uncertainty modelling
  • Observer-based techniques for fault detection and isolation
  • MATLAB/Simulink implementation
• Lecture 5 (Friday): Practical Session and Case Study Analysis (I)
  • Integrated model-based and data-driven methodologies
  • Integrating first week's topics: introduction to artificial intelligence tools
  • Discussion and preliminary feedback

Week 2

• Lecture 6 (Monday): Data-Driven Fault Detection Methods (I)
  • Introduction to data-driven techniques
  • Artificial intelligence approaches for anomaly detection
  • Machine learning techniques
  • MATLAB practical session
• Lecture 7 (Tuesday): Data-Driven Fault Detection Methods (II)
  • Machine learning approaches
  • Neural networks and Support Vector Machines (SVM)
  • Practical examples in MATLAB
• Lecture 8 (Wednesday): Advanced Data-Driven Approaches
  • Neural network techniques for fault detection
  • Signal processing methods and machine learning solutions
  • Hands-on exercises using MATLAB and Simulink
• Lecture 9 (Thursday): Integrated Model-Based and Data-Driven Approaches
  • Hybrid approaches
  • Comparative analysis and performance evaluation
  • Practical workshop and simulations
• Lecture 10 (Friday): Comprehensive Workshop and Final Discussion
  • Fault detection applications
  • Full integration of model-based and data-driven techniques
  • Open discussion, course wrap-up, and final Q&A session

Lecture General Topics

• Introduction: Course Introduction
• Issues in Model-Based Fault Diagnosis
• Fault Detection and Isolation (FDI) Methods based on Analytical Redundancy
• Model-based Fault Detection Methods

 
• Issues in Model-Based Fault Diagnosis
• Model Uncertainty and Fault Detection
• The Robustness Problem in Fault Detection
• System Identification for Robust FDI
• Fault Identification Methods
• Modelling of Faulty Systems

 
• Residual Generation Techniques
• The Residual Generation Problem

 
• Fault Diagnosis Technique Integration
• Neural Networks in Fault Diagnosis

 
• Output Observers for Robust Residual Generation
• Unknown Input Observer (UIO): Fundamentals
• FDI Schemes Based on UIO and Output Observers
• Kalman Filtering and FDI from Noisy Measurements: Fundamentals
• Residual Robustness to Disturbances

 
• Application Examples

Lecture Notes

• Part 1. Introduction to Fault Diagnosis, Residual Generation and Evaluation, Robustness Problems. PDF file.
• Part 2. Recursive Least Squares with Forgetting Factor for Detection of Process (System Component) Faults. PDF file.
• Part 3. Introduction to Artificial Neural Networks, Training and Validation. PDF file.
• Part 4. Dynamic ("Quasi-Static") Neural Networks for Fault Diagnosis. PDF file.
• Part 5. Fuzzy Logic and Fuzzy Models for Fault Diagnosis. PDF file.

Instructor Biography

Silvio Simani is a Professor of Automatic Control at the Department of Engineering, University of Ferrara, Italy. With over 25 years of academic and research experience in fault diagnosis, fault-tolerant control, and advanced control systems, Prof. Simani has authored numerous highly cited journal articles, international conference proceedings, and books in the fields of automatic control and aerospace engineering. He has extensive international teaching experience, including courses, summer schools, and specialised seminars worldwide.

Prof. Simani actively collaborates with renowned academic institutions, including Nanjing University of Aeronautics and Astronautics (NUAA), to promote international research projects and knowledge exchange among students and researchers worldwide.

For any questions or further information, please get in touch with Prof. Simani by email or WhatsApp at:
silvio.simani@unife.it
+39 320 304 4076.

Application Examples in Matlab and Simulink

• Example of System Modelling and Fault Topology. MATLAB script file with the process model; Simulink model to be discussed during the lecture; Final Simulink model implemented during the lecture.
• Example of Parametric Fault Generation in a Dynamic Process. Simulink file for the data generation with parametric changes.
• Example of Output Observer Design for a SISO Model and Estimation Error Generation. MATLAB script file with process model and observer design; Comprehensive Simulink model of process and observer for estimation error generation; Example model with noise and faults.
• Design Example of the Observer Bank for a SIMO Model (3 Dynamic Observers) for FDI. Matlab script file with the process model and the observer design; Simulink model with output sensor faults; Simulink model with input sensor faults; Matlab script file for residual evaluation.
• Recursive Estimation for the Diagnosis of Parametric Faults of a Dynamic Process. Implementation of the RLS Algorithm with Forgetting Factor; Matlab script for the visualisation of the results; Simulink file for the data generation with parametric changes; Simulink file for the generation of input-output faults.
• Recursive Estimation with Forgetting Factor and Parametric Faults of a Dynamic Process. Simulink file with RLS Simulink block (Matlab R2016).
• Comparisons of the Recursive Estimations in Simulink for Parametric Fault Diagnosis. Simulink file with the estimate generation via the RLS Simulink block.
• Learning Example for a Linear Perceptron. Training example in Matlab for a perceptron with linear separable patterns; training function for a linear perceptron in Matlab; Matlab Live Script example; binary classification example for a linear perceptron.
• Training Example of a Neural Network with Two Layers. Neural network training example in Matlab.
• Training Examples of a Neural Network with Back-Propagation. Matlab example of implementation of the back-propagation algorithm; simple implementation of the training algorithm in Matlab for a neural network; training data file.
• Use of the Neural Network Toolbox Example and Nonlinear Static Function Approximation. Data generation for the neural network training; Matlab script file for the training of an MLP neural network with Back-Propagation; Simulink file for the validation of the trained neural network (NN to be included); Matlab file for the verification of the trained network performance; Simulink file for the trained neural network (NN included); Simulink file for the neural network simulation with fault (NN and step fault).
• Example of Fuzzy c-means Algorithm for Fuzzy Clustering. Fuzzy c-means Matlab function; 2D fuzzy clustering example; 3D fuzzy clustering example; Zipped folder with the files (.zip file).
• Examples of Adaptive Neuro-Fuzzy Inference System - NN and ANFIS for Dynamic Process Modelling: Chemical Reactor Case Study. Compressed folder with files used in the simulations (.zip file).
• Modelling and Simulation of a Magnetic Levitation System for Fault Diagnosis. File with problem definition files (.zip file, 65kB); file with problem solutions (file .zip, 7MB).

References: Fundamental Monographs and Textbooks on FDI

• Rolf Isermann. "Fault-Diagnosis Applications: Model-Based Condition Monitoring: Actuators, Drives, Machinery, Plants, Sensors, and Fault-tolerant Systems". Springer. (April 29, 2011). ISBN: 978-3642127663.
• Steven X. Ding, "Model-based Fault Diagnosis Techniques: Design Schemes, Algorithms, and Tools". Springer, (April 10, 2008). ISBN: 978-3540763031.
• Rolf Isermann, "Fault-Diagnosis Systems: An Introduction from Fault Detection to Fault Tolerance". Springer-Verlag, 2005, 1st Edition. November, 28, 2005. ISBN: 3540241124.
• Blanke, M. and Kinnaert, M. and Lunze, J. and Staroswiecki, M. Schroder, J., "Diagnosis and Fault-Tolerant Control". Springer, 2003. 1st Edition. August 5, 2005. ISBN: 3540010564.
• Korbicz, J. and Koscielny, J. M. and Kowalczuk, Z. and Cholewa, W., "Fault Diagnosis: Models, Artificial Intelligence, Applications". Springer-Verlag, 2004. 1st Edition. February 12, 2004. ISBN: 3540407677.
• Simani, S. and Fantuzzi, C. and Patton, R. J., "Model-based fault diagnosis in dynamic systems using identification techniques", Springer-Verlag, 2002. ISBN 1852336854. Advances in Industrial Control Series. London, UK. First Eq. November, 2002. (298 pages).
• Basseville, M. and Nikiforov, I. V., "Detection of Abrupt Changes: Theory and Application", Springer-Verlag (March 1986), ISBN: 0387160434.
• Chen, J. and Patton, R. J., "Robust Model-Based Fault Diagnosis for Dynamic Systems", Kluwer Academic Publishers, 1999. ISBN: 0792384113.
• Chiang, L. H. and Russel, E. L. and Braatz, R. D., "Fault Detection and Diagnosis in Industrial Systems", Springer-Verlag London Limited, 2001. Advanced Textbooks in Control and Signal Processing Series. London, Great Britain. ISBN: 1852333278.
• Gertler, J., "Fault Detection and Diagnosis in Engineering Systems". Marcel Dekker, 1998, New York. ISBN: 0824794273.
• Hadjicostis, Christoforos N., "Coding Approaches to Fault Tolerance in Combinational and Dynamic Systems", Kluwer Academic Publishers. November 2001. The Kluwer International Series in Engineering and Computer Science. ISBN: 0792376242.
• Liu, G. P. and Patton, R. J., "Eigenstructure Assignment for Control System Design", John Wiley and Sons. England, 1998. ISBN: 0471975494.
• Patton, R. J. and Frank, P. M. and Clark, R. N., "Fault Diagnosis in Dynamic Systems, Theory and Application", Prentice Hall. 1989, London. Control Engineering Series. ISBN: 0133082636.
• Patton, R. J. and Frank, P. M. and Clark, R. N., "Issues of Fault Diagnosis for Dynamic Systems", Springer-Verlag, 2000. London Limited. ISBN: 3540199683.

Downloads: Application Examples

• "Model-Based Fault Diagnosis for Industrial Processes" (Silvio Simani's Extended Report, October 2007): (PDF file, 35 MB).
• "Lecture Notes, Chapters 1 and 2" (Chapters from Silvio Simani's Extended Report, October 2007): (PDF file, 1 MB).
• "Model-based fault-detection and diagnosis - status and applications" (Journal Paper by Rolf Isermann, 2005): (PDF file, 1 MB).
• Design Example of Output Observer for FDI. Example with Noise (Matlab and Simulink files and models for Matlab 6.1): (zipped Matlab and Simulink files, 7 KB).
• Design Example of Output Observers for FDI. SIMO Model with three Observers. (2 Matlab files and 1 Simulink model for Matlab 6.1): (zipped Matlab and Simulink files, 5 KB).
• Design Example of a Kalman filter. Model with noise errors. Generation of minimal variance estimation error signals. (Matlab and Simulink files for Matlab 6.1): (zipped Matlab and Simulink files, 5 KB).
• Kalman filter for Fault Diagnosis. Model with noise errors and output sensor fault. Residual statistical tests. (Matlab and Simulink files for Matlab 6.1): (zipped Matlab and Simulink files, 7 KB).
• Examples of nonlinear models and neural network training. Zipped Matlab and Simulink directories (14 MB).
• Examples of integration of neural networks and fuzzy models with dynamic observers and filters for fault detection and isolation. Zipped Matlab and Simulink directories (5 MB).

Previous Page

Simani Home Page

Dipartimento di Ingegneria