Abstract

Large-scale dynamical systems attract a significant and steadily growing interest in academia and industry. Examples of these systems include power and water distribution networks, smart buildings, complex industrial plants and transportation systems. They are characterized by a large number of states and inputs, are generally spatially distributed, and are modeled as the interaction of many subsystems coupled through physical or communication relationships. They can fail due to component faults, human errors, natural disasters or intentional malicious attacks. Reliability is a key requirement in systems design and the development of distributed methods for fault diagnosis and fault-tolerant control is an emergent important topic. Furthermore, Large-Scale Systems can often have a dynamic structure that changes over time: the topology of the network of interconnected systems may evolve, or the revamping of some subsystems may change the relationships with the rest of the network, or the occurrence of a fault may cause the disconnection of a subsystem in order to preserve safety and reliability of the rest of the system. In this talk, I will present a distributed model-based fault detection and isolation approach and I will explain how to use it in a scalable fault-tolerant control scheme as a decision support tool for the reconfiguration of the control.