An analytical tuning approach to multivariable model predictive controllers

• Analytical tuning approach to multivariable model predictive controllers.
• Multivariable first-order plus dead time models.
• Closed loop decoupling conditions.
• Achievable and feasible performances of closed loop.
• Simulation results available for multivariable pH process and Wood and Berry plant.

Multivariable model predictive control is a widely used advanced process control methodology, where handling delays and constraints are its key features. However, successful implementation of model predictive control requires an appropriate tuning of the controller parameters. This paper proposes an analytical tuning approach to multivariable model predictive controllers. The considered multivariable plants are square and consist of first-order plus dead time transfer functions. Most of the existing model predictive control tuning methods are based on trial and error or numerical approaches. In the case of no active constraints, closed loop transfer function matrices are derived and decoupling conditions are addressed. For control horizon of one, analytical tuning equations and achievable performances are obtained. Finally, simulation results are used to verify the effectiveness of the proposed tuning strategy.