Distributed Diagnosis on DES

Diagnosis on Hybrid Systems

Overview of my PhD work and experience

The complexity of modern-world systems, like the electricity network, is a challenge to manage. I am part of the project AI for the smart grid at National ICT Australia (NICTA), whose main aim is to develop technologies for automating the management of an enhanced electricity grid overlayed with an information system, sensors and control devices (a smart grid). The desired goal is to achieve robustness and fault tolerance in the network.

One critical aspect to achieving such a goal is the ability to diagnose system behaviour. We want the system to operate as smoothly and efficiently as possible until faults are repaired. Knowing what happened on the system is a key factor to achieving that. We define diagnosis as the problem of determining what happened on a system given observations made on the system.

My work is concerned with the diagnosis of large complex distributed systems, with electricity networks being the focus application. I use a model-based approach where it is assumed that a model of the system is available. System behaviour is captured by automata which are one way to mathematically model finite state machines. This allows us to capture the different modes the system can be in and transitions that occur between those modes as various events take place in the system. For example, in an electricity network, a generator can be in a number of operational modes (e.g. providing power or providing spinning reserve). These modes are discrete and we can use a discrete-event system representation to describe system evolution where mode transition is triggered by the occurrence of one or more events.

Next