A convex approach for NMPC based on second order Volterra series models

This paper presents a novel approach to use second order Volterra series models in nonlinear model predictive control. A common technique in model predictive control is the minimization of a quadratic cost function with respect to the future input sequence. In the case of nonlinear models, the resulting cost function is a possibly non-convex function. The proposed strategy uses quadratic cost functions to approximate the original cost function. For the quadratic cost functions, convexity can be achieved easily by adding a weighting function of the control increments. The approximated convex cost functions are minimized globally by means of an iterative approach with guaranteed convergence. The proposed control strategy is applied to a continuous stirred tank reactor and the control performance is illustrated by experimental results.