Comparative analysis of sliding mode and Volterra model predictive controllers applied to a benchmark CSTR non-minimum phase reactor

The success of linear model predictive control in industry has not been replicated in the non linear formulation. This paper presents two different ways of approaching NMPC techniques to the industry use. On the one hand the Sliding Mode Predictive Control (SMPC) blends the simplicity of linear MPC with sliding mode control techniques, using for prediction a linear approximation of the process and enhancing its performance with the sliding formulation. On the other hand Volterra Model based Predictive Controllers (V-MPC) offer a non linear predictive controller that uses a time efficient algorithm to solve the optimization problem. The performance of both controllers was compared using a non minimum phase isothermal Van de Vusse reactor process, and a linear MPC was used as a reference of performance for the class of linear controllers.