Page 32 - Distributed model predictive control for plant-wide systems

6 Distributed Model Predictive Control for Plant-Wide Systems

sp sp sp sp sp
y 1 y 2 y 3 y m−1 y m

Decentralized
Cont 1 Cont 2 Cont 3 Cont m-1 Cont m controllers

*
* y m
u 1 y 1 u 2 * y 2 u 3 * y 3 * u m–1 y m–1 u m
S *
Distributed
S Na system
S 1 S 2
S *
S 3 S Na-1

Figure 1.6 Decentralized control

work due to some fault, the other controllers are still able to work, which means good error
tolerance. Furthermore, if there are some new subsystems required to be appended to or deleted
from the existing plant-wide system, it needs to do nothing with the existing controllers, which
means good structure flexibility.
However, since there is no communication and coordination among decentralized con-
trollers, the controller performance is destroyed if the coupling among subsystems is strong
enough. In order to avoid the degradation of the performance of the global system, one method
is to enlarge the scale of each local controller, where several strong coupled subsystems are
controlled by one local controller. By using this strategy, the performance of the global system
could be guaranteed, but the computational burden of each local controller is increased, and
the flexibility of overall system is deduced. This strategy bypasses and does not solve the
problem of how to improve the global performance when strong interactions exist among the
subsystems each of which is controlled by a separated controller.
Unfortunately, in most cases, strong couplings exist in the plant-wide system. Thus, peo-
ple add a coordinator to coordinate each subsystem-based controller for improving the global
performance of the entire plant-wide system, as shown in Figure 1.7, and is called hierarchi-
cal coordinating decentralized control. Through different coordinating algorithms, the global
performance of the entire system could be significantly improved if strong interactions exist.
However, all local controllers should communicate with the coordinator as the global infor-
mation is necessary for the coordinator. The centralized structure appeared in the coordinator.

1.2.3 Distributed Control
Recently, with the development of computer technologies, fieldbus, network communicating
technologies, and smart meter in process industries, which allows the control technologies and
methodologies to utilize their potentials for improving control, the distributed control structure
has appeared gradually instead of the centralized and decentralized structure for the plant-wide
system. As shown in Figure 1.8, the global system is divided into many interacted subsystems,
and each subsystem is controlled by a separate controller; these peer controllers communicate