| Motivation and rationale |
Besides
having a high methodological importance, event-based control is
nowadays viewed in the industry as a means to reduce traffic on control
networks, and also to avoid frequent small actuator moves that are
irrelevant for the plant, but detrimental for the actuator's lifetime.
This interest is enhanced by an increasing one in wireless control
systems, but the importance of the matter is general. My research in this context is quite recent and basically application-driven, though also exhibiting some methodological aspects. In one word, the rationale is to abstract the widest possible set of characteristics that an industrial implementation prescribes to a system, take those as hypotheses in modelling and control synthesis, and exploit this generality sacrifice to viably guarantee properties that would be far more difficult to assess in the general case. |
| Methodological achievements |
The
research naturally lends themselves to the periodic event-based control
setting, and to date refers to loops with a single source of events.
Such a context naturally induces in the control system a switching
nature, as - simplifying for brevity - at a certain step the control
signal is either updated or held. In the case of sensor-generated events, the idea was proposed to have the sensor transmit past samples of the controlled variable, at its internal fixed sampling rate, with respect to the one that generated the event. Based on this additional information, sufficient but simple (and in general reasonably conservative) guarantees can be achieved on closed-loop stability. |
| Technological achievements |
Event-based
control has been crucial for the results obtained in
thermal/power/performance management for microprocessors. Besides this,
the developed techniques were applied in several studies, and are
forming the basis for a library of control blocks specified in a
functional manner, so as to be included in industrial control systems. |
| Open issues and work in progress |
On
the methodological side, relationships are being established with the
automatic tuning of controllers, to exploit the additional parameters
of an event-based realisation (such as event triggering thresholds,
minimum inter-event time, and so on) as part of the tuning process
itself. Studies are also underway on the role of the hold mechanism. On the technological side, the ideas above are being used to develop a comprehensive model and software library for event-based control. |
| Outlook and vision |
The
industrial interest on event-based control is expected to increase, and
the hope is that the proposed solutions - that have a strong
application grounding - encounter success. Besides and very important, the ability of realising a control solution in event-based form is extremely useful in the context of computing systems. At low levels - i.e., near to the hardware - this is important to reduce overheads; at high levels, like in a distributed application, it helps mitigate the control-generated traffic. Plans are to deepen, prove with studies, and disseminate knowledge on these aspects, that are potentially important drivers for the success of a control-centric design and management approach. |