Plug and play fault-tolerant control
Lead Research Organisation:
Imperial College London
Department Name: Electrical and Electronic Engineering
Abstract
This project focusses on the development of distributed and Plug-and-Play monitoring techniques for fault tolerance and security within Large Scale and Cyber Physical Systems, with a particular focus to DC microgrids.
Modern day engineering systems, due to their large scale, special sparseness, and physical and cyber interconnections, are often studied as Large Scale Systems (LSS) or Cyber Physical Systems (CPS). These categories include many of the infrastructures which are critical to modern day life, such as water treatment facilities, and energy transmission and distribution networks. Because of their centrality, it is crucial to ensure the safety and security of these systems, and therefore to be able to detect any anomalies, which may be caused by a fault in a component, or an attack by a malicious agent on the network.
Because of their scale and interconnected nature, LSS and CPS cannot be monitored with centralized monitoring techniques, as they require communication and computational resources, which may be infeasible or too costly. Hence, distributed architectures must be designed to monitor the health of the system.
This project addresses the need for a distributed monitoring strategy, with the intent of developing not only anomaly detection layers for control systems, but to also define a methodology of reconfiguring the control once an anomaly has been detected.
In addition to its distributed architecture, the developed methods aim to have Plug-and-Play capabilities. Plug-and-Play defines the property of controller design by which if the structure of the LSS/CPS changes, i.e. a subsystem is added or removed, only a subset of the controllers must be re- tuned to account for the change.
The proposed techniques are then demonstrated on a linear time invariant model of an islanded DC microgrid, an energy distribution network in which generation and consumption of electricity is collocated within a single node, which has gained attention in literature due to its possible use for the integration of renewable energy sources within energy distribution.
This project is aligned with the EPSRC research area of Control Engineering.
Modern day engineering systems, due to their large scale, special sparseness, and physical and cyber interconnections, are often studied as Large Scale Systems (LSS) or Cyber Physical Systems (CPS). These categories include many of the infrastructures which are critical to modern day life, such as water treatment facilities, and energy transmission and distribution networks. Because of their centrality, it is crucial to ensure the safety and security of these systems, and therefore to be able to detect any anomalies, which may be caused by a fault in a component, or an attack by a malicious agent on the network.
Because of their scale and interconnected nature, LSS and CPS cannot be monitored with centralized monitoring techniques, as they require communication and computational resources, which may be infeasible or too costly. Hence, distributed architectures must be designed to monitor the health of the system.
This project addresses the need for a distributed monitoring strategy, with the intent of developing not only anomaly detection layers for control systems, but to also define a methodology of reconfiguring the control once an anomaly has been detected.
In addition to its distributed architecture, the developed methods aim to have Plug-and-Play capabilities. Plug-and-Play defines the property of controller design by which if the structure of the LSS/CPS changes, i.e. a subsystem is added or removed, only a subset of the controllers must be re- tuned to account for the change.
The proposed techniques are then demonstrated on a linear time invariant model of an islanded DC microgrid, an energy distribution network in which generation and consumption of electricity is collocated within a single node, which has gained attention in literature due to its possible use for the integration of renewable energy sources within energy distribution.
This project is aligned with the EPSRC research area of Control Engineering.
Organisations
People |
ORCID iD |
| Thomas Parisini (Primary Supervisor) | |
| Alexander Gallo (Student) |
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/N509486/1 | 01/10/2016 | 31/03/2022 | |||
| 1859613 | Studentship | EP/N509486/1 | 01/10/2016 | 31/03/2020 | Alexander Gallo |
| Description | A distributed architecture has been developed for the cyber security of large-scale systems, with a specific focus towards the application of DC microgrids, i.e. an energy generation and distribution network which may be key in the future power grid, as it is capable of efficiently interfacing multiple generation and consumption nodes. In the operations of large-scale systems, other than fault-tolerant control, it is fundamental to address the possibility of malicious attacks within the communication infrastructure used for normal operations. Methods for attack detection can be developed following some suggestions present in the literature for fault detection and isolation, and further work must be done towards attack resilience. The problem of attack detection in communication networks has been addressed, and has resulted in a number of publications, including a submission towards the Special Issue of the IEEE Transactions of Automatic Control, "Security and Privacy of Distributed Algorithms and Network Systems". The architecture proposed has been proven effective in a realistic numerical simulation setting. Furthermore, it has been designed to be scalable with the size of the network, thus providing an initial framework for it to be plug-and-play, i.e. for it to automatically handle subsystems "entering" and "leaving" the network. |
| Exploitation Route | Following detection, a reconfiguration strategy is to be proposed, allowing the large-scale system to automatically adapt to the presence of an attack. Furthermore, the integration of fault isolation capabilities into the architecture may be considered. The proposed cyber-attack detection architecture will be tested on a physical testbench, providing crucial insights on its applicability in a real world setting. |
| Sectors | Aerospace, Defence and Marine,Digital/Communication/Information Technologies (including Software),Energy,Security and Diplomacy |