SEMINAR: A neural network-based fault-tolerant control

Guest: Marcin Mrugalski, University of Zielona Góra

Title: A neural network-based fault-tolerant control (EE, CS, DS, ME, MFG)

Date/Time: January 7, 2026, 13:40

Location: FENS G035

Abstract: The presentation is devoted to the problem of designing a robust sensor and actuator faults estimator based on a recurrent neural network by using a linear parameter-varying form. In particular, basic concepts in the field of diagnostics and fault-tolerant control in technical systems are presented. The process of designing and analysing the convergence is carried out by the quadratic boundedness approach. Based on the obtained estimator, a fault-tolerant controller for multiple sensor and actuator faults was designed. The developed approach reduces the impact of simultaneous sensor and actuator faults on the controlled system. The robust estimator and controller design procedure boils down to solving a set of linear matrix inequalities. The effectiveness of the developed fault-tolerant control method is shown on a laboratory multi-tank system. In particular, the results of control in the event of sensor failure and a drop in the effectiveness of the pump supplying the tank are presented.

Bio: Prof. Marcin Mrugalski received the Ph.D. degree in electrical engineering from the University of Zielona Góra, Poland in 2004. He obtained the D.Sc. degree in computer science from the Częstochowa University of Technology in 2014. He has been an associate professor of automatic control and robotics at the Institute of Control and Computation Engineering, University of Zielona Góra, since 2014. Since 2015, he is a member of the Control Engineering and Robotics Commission of the Polish Academy of Sciences. His current research interests include analytical methods, neural networks, fuzzy logic, computational intelligence, parameters estimation and their applications in fault diagnosis, fault-tolerant control and design of experiments. Another trend in research is the use of IT techniques in scheduling and monitoring production and transport processes. He was the project manager or a member of research teams under 11 grants, financed by the European Commission, the National Science Centre (NSC), the Ministry of Science and Higher Education (MSHE), and INTERREG. He was the project manager of the "University Patent Centre" project funded by NSC, which resulted in the submission of 32 applications to the European Patent Office. He is currently the manager of a research task in the project “MIST: Artificial intelligence-assisted design of experiments as a tool to streamline the development of siRNA-lipid nanoparticle therapeutics in soft mist inhalers”, funded by NSC. He has published one monograph and more than 80 papers in international journals, book chapters, conference proceedings, and is a co-author of one European Patent. His teaching activities focus on computer networks, VoIP telephony, and network security. For 20 years, he has been the manager and instructor of the Cisco Networking Academy at the University of Zielona Góra.