A new computationally efficient formulation of non-linear predictive controller for dynamically rapid industrial applications
Loading...
Files
Date
2018
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of New Brunswick
Abstract
As the computational power available in the control loop increases, the use of model predictive controllers are becoming more cost effective. This has allowed them to be used in industrial facilities but only on certain systems. Systems with fast, non-linear dynamics remain out of reach due to the computational requirements every control cycle. Through leveraging the seminal dynamic matrix model predictive control scheme along with discrete time modeling techniques, an alternative formulation has been developed. Through investigation of the premise that a model predictive controller could be reduced the theory for a stateless discrete predictive controller is derived. Through a series of simulation and experiments the performance of this new controller structure was evaluated. This allowed the theoretical framework of the controllers to be validated in practice. As designed into the controller, the most significant improvement is in the computational time required each control cycle. In addition, the theoretical formulations demonstrated that the computational time can be dynamically fixed irrespective of changing controller parameters. This feature in itself is advantageous as it allows unusually long prediction horizons to improve stability, which was not previously possible.
The stateless discrete predictive controller also has the advantage of being stateless, in the sense that it does not base its current prediction on any previous predictions. This allows the controller to adjust quickly to model mismatch or disturbances the system encounters. The discrete formulation of this predictive controller allows for the stateless prediction to be performed without compromising computational time. The simulation and practical results demonstrated the superior control performance in comparison to standard predictive control.