SYSTOL 2016

  • Increase font size
  • Default font size
  • Decrease font size
Sponsors: MCA UPC MINECO ERDF Technical
co-sponsors:
IEEE CSS RAS CEA-IFAC

Plenary session: Thomas Parisini

A Distributed Approach for Plug and Play Monitoring and Fault-tolerant Control of Large-scale Systems


This lecture deals with a class of systems that are becoming ubiquitous in the current and future "distributed world" made by countless "nodes", which can be cities, computers, people, etc., and interconnected by a dense web of transportation, communication, or social ties. The term "network", describing such a collection of nodes and links, nowadays has become commonplace thanks to our extensive reliance on "connections of interdependent systems" in our everyday life, for building complex technical systems, infrastructures and so on. In an increasingly "smarter" planet, it is expected that such interconnected systems will be safe, reliable, available 24/7, and of low-cost maintenance. Therefore, health monitoring, fault diagnosis and fault-tolerant control are of customary importance to ensure high levels of safety, performance, reliability, dependability, and availability. For example, in the case of industrial plants, faults and malfunctions can result in off-specification production, increased operating costs, production line shutdown, danger conditions for humans, detrimental environmental impact, and so on. Faults and malfunctions need to be detected promptly and their source and severity should be diagnosed so that corrective actions can be taken as soon as possible. Once a fault is detected, the faulty subsystem can be unplugged to avoid the propagation of the fault in the interconnected large-scale system. Analogously, once the issue has been solved, the disconnected subsystem can be re-plugged-in.

In the talk, an adaptive approximation-based distributed fault diagnosis approach for large-scale nonlinear systems will be dealt with, by exploiting a "divide et impera" approach in which the overall diagnosis problem is decomposed into smaller sub-problems, which can be solved within “local” computation architectures. The distributed detection, isolation and identification task is broken down and assigned to a network of "Local Diagnostic Units", each having a "local view" of the system.

Moreover, the lecture will address the integration of a distributed model predictive control scheme and a distributed fault diagnosis architecture. Specifically, in the off-line control design phase we adopt a decentralized algorithm and we assume that the design of a local controller can use information at most from parents of the corresponding subsystem, i.e., subsystems that influence its dynamics. This implies that the whole model of the large-scale system is never used in any step of the design process. This approach has several advantages in terms of scalability: i) the communication flow at the design phase has the same topology of the coupling graph - usually sparse - ii) the local design of controllers and fault detectors can be conducted independently; iii) local design complexity scales with the number of parent subsystems only; iv) if a subsystem joins/leaves an existing network (plug-in/unplugging operation) at most children/parents subsystems have to retune their controllers and fault detectors. We refer to this kind of decentralized synthesis as plug & play design, if - in addition - the plug-in and unplugging operations can be performed through a procedure for automatically assessing whether the operation does not spoil stability and constraint satisfaction for the overall large-scale system.