I

Marie Curie Intra European Fellowhip (Project Number: 626891)

`FUTuRISM: Multiple sensor FaUlt ToleRant control for management of Interconnected nonlinear SysteMs'

RESEARCH OBJECTIVES:

The primary research objective of this project is the design and analysis of novel methods for diagnosing multiple sensor faults and compensating their effects on multi-sensory schemes used for controlling interconnected, nonlinear systems. The second main objective of this project is the application of these methods to complex systems. To this end, the four pillars of this research are:

I. Multiple SFD with detection and isolation guarantees for nonlinear systems

In order to develop a methodology for multiple SFD in multi-sensor controlled nonlinear systems, enhanced with detection and isolation guarantees, this research will focus on:

  1. Designing a set-theoretic multiple SFD methodology for nonlinear systems: Existing methodologies for detecting and isolating multiple sensor faults in nonlinear systems are developed using the open-loop system model (see Figure 1(a)). However, the substandard performance of a multiple SFD methodology can have a signicant impact on the effectiveness of an active, multi-sensor FTC scheme. Thus, there is a need for guaranteeing multiple sensor fault detection and isolation. These guarantees can be obtained by taking into account the closed-loop system operation and deriving positive invariant sets. Although there are well-established set-theoretic methods for computing positive invariant sets for linear systems, there are very few for nonlinear systems. This research will elaborate on developing a multiple SFD technique for multisensory-controlled, nonlinear systems, exploiting the information stemming from the closed-loop operation (see Figure 1(b)), while computing positive invariant sets for nonlinear systems.
  2. Analyzing the multiple SFD methodology based on positive invariant sets: The state of the art in multiple SFD for nonlinear systems provides certain conditions for fault detectability and isolability taking into account the open-loop system operation. This research will  roceed to (i) analyze the detectability and isolability of the developed multiple SFD methodology with respect to multisensory-controlled operation and general class of sensor faults, and (ii) derive multiple sensor fault detection and isolation guarantees.

II. Multiple sensor FTC with fault tolerance guarantees for nonlinear systems

The second goal of this research will be to investigate active fault tolerant control approaches for compensating the effects of multiple sensor faults that occur in nonlinear systems, along the following lines:

  1. Design of multiple sensor FTC scheme for nonlinear systems: Active FTC schemes, which rely on switching between or fusing observer-based state estimates, or virtual sensors, are well-established for linear systems under structural assumption on the dynamics (principally linearity) and restrictive (single-sensor) fault scenarios. Based on this assumption, these FTC schemes can provide fault tolerance guarantees using positive invariant sets. Moreover, although these FTC schemes use multi-sensor systems, they do not usually exploit any information related to the type or distribution of the sensors. This research will focus on designing FTC schemes that incorporate a multiple SFD mechanism, giving emphasis to the control reconguration for handling multiple sensor faults in nonlinear systems, while investigating the use of spatial and temporal correlations in the design of these schemes. 
  2. Analysis of multiple sensor FTC schemes: The design of an active FTC scheme should be analyzed in terms of stability and tracking performance. To this end, during this research we will analyze the stability and tracking performance ensured by the developed multiple sensor FTC schemes and derive fault tolerance guarantees, taking into account modeling uncertainty and multiple sensor faults.

​III. Multiple sensor FTC of interconnected, nonlinear systems

The third research objective will be the design of non-centralized, multi-sensor FTC schemes for interconnected, nonlinear systems, taking into account practical issues related to the imperfections of the network communication between the computing nodes. Thus, this research will proceed along the following lines:

  1. Design of non-centralized multiple sensor FTC architectures for interconnected nonlinear systems: The existing techniques for developing an active FTC for multi-sensor systems with fault detection and isolation guarantees, as well as fault tolerance guarantees, have been deployed following a centralized approach. On the other hand, sensor FTC schemes that can handle multiple sensor faults in interconnected systems are very limited. In this research project, we will elaborate on designing non-centralized architectures for the deployment of multiple sensor FTC mechanisms in order to compensate the effects of multiple sensor faults that may occur in more than one interconnected subsystems of a large-scale, nonlinear system. 
  2. Management of multiple sensor FTC schemes under communication imperfections: In large-scale, interconnected systems, communication between various physical and cyber components in the loop is usually carried out with time-delays, such as network access and transmission delays, and packet dropouts due to network congestion or transmission errors in physical links. Both time delay and packet dropouts may have a signicant effect on the performance of non-centralized, active FTC schemes. Therefore, this research will focus on the management of non-centralized, multiple sensor FTC schemes in relation to communication problems in a time-delay dynamical systems modeling framework. 
  3. Analysis of non-centralized multiple sensor FTC architectures: During the design of the multiple sensor FTC architectures for interconnected, nonlinear systems, we will analyze the performance of the developed architectures with respect to stability and tracking performance, both locally and globally, taking into account communication problems.

IV. Application of multiple sensor FTC to complex systems

We will aim at applying the developed SFD and FTC techniques to complex systems, which are nonlinear and may consist of a large number of subsystems, while multiple sensors are used for their monitoring and control. Real-world examples of such complex systems are the intelligent transportation systems and smart buildings systems. The intelligent transportation systems are information and communication technologies (ICT) used in the eld of road transport, aiming at ensuring road and vehicle safety. Towards this goal, a large number of sensors is used for cooperative and autonomous driving, e.g. for controlling the lane departure of a vehicle or for keeping safe distance in a vehicle platoon. A smart building is a building that incorporates computer technology in order to increase energy effciency, improve comfort, productivity and safety. Multiple sensors, installed in every zone (rooms, corridors etc) and the Heating Ventilation Air-Conditioning (HVAC) system, are needed for monitoring and control of the energy and the living conditions of a large-scale smart building.

CENTRALESUPELEC: 

  • Sorin OlaruScientist in charge, Professor, Laboratoire des Signaux et Systèmes (UMR CNRS 8506), CentraleSupélec, Gif-Sur-Yvette, France
  • Abid Rahman Kodakkadan, PhD Student

EXTERNAL COLLABORATORS: 

KIOS Research Center for Intelligent Systems and Networks, Dept. of Electrical and Computer Engineering, University of Cyprus, Nicosia, Cyprus

Department of Automatic Control and Systems Engineering"Politehnica" University of Bucharest, Bucharest, Romania

 

RELATED PROJECTS:

PHC GALILEO: "Set theoretic analysis of switched and time delay systems with application to fault tolerant control systems"

Partners, University of Udine, Italy

Partners, CentraleSupélec France

PHC PESSOA: "Robust Distributed Model Predictive Control of Medium- and Large- Scale Systems"

Partners, University of Porto, Portugal

Partners, CentraleSupélec, France