Final Report Summary - FAULT-ADAPTIVE (Fault-Adaptive Monitoring and Control of Complex Distributed Dynamical Systems)
Modern society relies on the availability and smooth operation of complex engineering systems. Examples include electric power systems, smart buildings, water distributions networks, transportation systems, etc. To facilitate monitoring and control of such systems new sensor/actuator devices are continually being developed at reduced costs and in larger quantities, while traditional monitoring and control specifications and objectives are being expanded to include additional requirements such as energy efficiency, environmental impact and security. In situations where a fault arises in some of the components (e.g. sensors, actuators, communication links), or an unexpected event occurs in the environment, this may lead to a serious degradation in performance or, even worse, to an overall system failure. Standard approaches are typically not able to handle abrupt significant changes in the dynamics due to a fault or persistent erroneous sensor data, while in some cases the feedback controller may contribute to “hiding” incipient faults that develop slowly over time, until it is too late to prevent a serious system failure.
The FAULT-ADAPTIVE project has developed an innovative monitoring and control framework with learning algorithms for approximating, during operation, key correlations between measured variables.
The developed framework is based on a hierarchical fault diagnosis architecture, with neighboring fault diagnosis agents cooperating at a local level, while transmitting their information, as needed, to a regional monitoring agent, responsible for integrating in real-time local information into a large-scale “picture” of the health of the underlying system. A key motivation was to exploit spatial and temporal correlations between measured variables using learning methods, and to develop the tools and design methodologies that will prevent relatively “small” faults or unexpected events from causing significant disruption or complete system failures in complex distributed dynamical systems. The work developed in this project has made significant contributions to the application of advanced fault diagnosis methods to critical infrastructure systems, such as power and energy systems, water distribution networks, and intelligent transportation systems.
The FAULT-ADAPTIVE project has developed an innovative monitoring and control framework with learning algorithms for approximating, during operation, key correlations between measured variables.
The developed framework is based on a hierarchical fault diagnosis architecture, with neighboring fault diagnosis agents cooperating at a local level, while transmitting their information, as needed, to a regional monitoring agent, responsible for integrating in real-time local information into a large-scale “picture” of the health of the underlying system. A key motivation was to exploit spatial and temporal correlations between measured variables using learning methods, and to develop the tools and design methodologies that will prevent relatively “small” faults or unexpected events from causing significant disruption or complete system failures in complex distributed dynamical systems. The work developed in this project has made significant contributions to the application of advanced fault diagnosis methods to critical infrastructure systems, such as power and energy systems, water distribution networks, and intelligent transportation systems.