Autonomic decentralized schedule execution

[Pieter Vansteenwegen]

for Dirk Cattrysse and Paul Valckenaers:

Scheduling systems account well for the 'big picture' when optimizing production, but they are less able to account for details.
Manufacturing execution systems manage production in detail but only account for local information when taking decisions.

Holonic Manufacturing execution systems are able to coordinate both locally and non-locally in a fully decentralized manner.
However, these holonic decentralized mechanisms are inefficient regarding global optimization relative to scheduling systems.
They propagate information but never use a 'global/aggregated picture.'

The thesis investigates how externally provided schedules may 'guide' the search and decision-making processes in this
Holonic MES. The basic scheme has been proposed in the forthcoming doctorate of Paul Verstraete (cf. below).

The topic of the doctoral thesis is to use and enhance this scheme:

* To investigate the mechanisms both in the scheduler and the Holonic MES, and their cooperation
* To study and optimize the (joint/overall) performance/robustness of this combination
* To establish 'good practice' in this autonomic schedule execution design.

Reference:

* Verstraete, P., Valckenaers, P., Van Brussel, H., Saint Germain, B., Hadeli, K., Van Belle, J. (2008).
Towards robust and efficient planning execution. Engineering Applications of Artificial Intelligence, 21(3), 304-314.