|SPINELLI STEFANO||Cycle: XXXII |
Section: Systems and Control
Tutor: GARATTI SIMONE
Advisor: FARINA MARCELLO Major Research topic
:Distributed Model Predictive Control for large-scale networked systems with shared resources Abstract:
The objective of the thesis is the study and the application of a novel algorithm for optimal management of large-scale networked systems, sharing a set of common resources.
The configuration and the dimension of the problem intrinsically pose the main issue of its tractability with standard centralized approaches. Therefore, the distribution of control intelligence is the key point to reach a plant-wide dynamic optimal control.
This class of problems embraces an important set of typical industrial systems. Application examples are:
- Multi-line chemical plants or parallel manufacturing processes with raw material constraints;
- Multi-stage production plants with energy/power constraints;
- Water/steam supplier/consumer networks.
These large-scale systems are composed of a set of dynamic components. Every subsystem is provided with a local controller based on a selfish control objective: the combination of the subsystem and its controller define an agent. Due to agent interactions and the presence of the shared resource coupling, the application of a pure decentralized approach has severe issues.
The study of a cooperative distributed model predictive control strategy will provide a coordination of these competitive agents. The subsystem feedback control is obtained as the solution to an optimization problem in which the information about other agent strategies is provided by an inter-controller communication.
The main issue of the control scheme studied in this thesis is the fact that complicating (i.e., coupling) constraints, caused by the sharing of a set of limited resources, must be enforced.
The first case, studied in the thesis, will consist of a network of subsystems that are the consumers of a shared resource. An extension of this case study consists of dealing, not only with consumers, but also with producers of such unitary resource.
A series of examples will show the applicability of this approach. Moreover, to demonstrate the performance of the proposed control strategy, a steam plant is considered as a use case. In this scenario, all the autonomous controllers of the generator units and the consumer systems will be replaced by local controllers, coordinated in a cooperative fashion by a suitable distributed control scheme to reach a more efficient operational point.