Decentralized Model Predictive Control Based On I/O Plant Model
The paper deals with the decentralized model predictive controller design. The unconstrained GPC is considered to design local controllers within the Equivalent subsystems methodology (ESM). According to ESM, the original multivariable plant is diagonalized by generating so called equivalent subsystems, for which local controllers are tuned independently. Resulting local controllers constitute the decentralized controller and are implemented on the real plant. Plant-wide closed-loop stability and performance under the decentralized controller are guaranteed if local controllers provide stability and required performance of equivalent subsystems.
The proposed approach has been verified on a case study - decentralized GPC design for a two-input two-output laboratory plant.