Project title: Industrial Embedded Model Predictive Control
Application domain: Automotive/Industrial Automation
Supervisors: Dr. Daniele Bernardini, Prof. Alberto Bemporad
Researcher in charge: Nilay Saraf
Project description:
Model Predictive Control (MPC) is widely spreading in industry as one of the most powerful techniques to design multivariable controllers that optimise closed-loop response under constraints
on system variables. The MPC algorithm solves a mathematical programming problem in real-time based on a dynamic model of the physical system.
The objectives of the project are
-To incorporate and deal with resource constraints such as memory, numerical precision and stability, etc. in the optimization based control design
-Develop MPC-oriented system identification toolchain
-Validate the developed tools and methods on real industrial test cases
The main contribution expected through this project is bridging the gap between state-of-the-art MPC theory and industrial practice, developing tools for production-intent MPC software that reduces time to market of new products.
Research abstract
Current research focuses on developing fast and efficient MPC algorithms for real-time applications. Fast MPC has been one of the key topics in research in order to deploy this advanced control technique in the industry on embedded platforms. This project investigates methods and algorithms through which this objective can be achieved. The methods developed are applicable to linear dynamical systems such that they can be extended for fast MPC of nonlinear and adaptive systems, which are more relevant to industrial applications.
Publications:
Fast model predictive control based on linear input/output models
and bounded-variable least squares, 56th IEEE Conference on Decision and Control, Melbourne
