Multi Model Adaptive Industrial MPC Controller

D. Thiele and W. Wojsznis (USA)


Adaptive Control, MPC, PID, DMC, DCS


This paper presents an adaptive MPC controller using multi-model evaluation. The adaptation uses a set of models, switching strategy, and parameter interpolation. Interpolation is based on the integrated squared error assigned to every value of the model parameter. The adaptation of model parameters is performed sequentially, which results in fast evaluation of multiple models and improved convergence. The first order plus dead time models are applied. The adapted parametric model is used for generating step response applied in DMC-type controller. The controller is developed in a form that adaptation does not require matrix operations. This feature along with computing requirements comparable with PID controller makes possible MPC application on a basic control level as an alternative to PID. The adaptive MPC controller demonstrates substantial advantages as compared to the PID control – an excellent response on set point changes for a wide range of process dynamics, comparable response on disturbances, more tolerance on adaptive model inaccuracy and additional functionality including split range and simple optimization.

Important Links:

Go Back