Skip to Main Content
Almost all industrial processes exhibit nonlinear dynamics, however most model predictive control (MPC) applications are based on linear models. Linear models do not always give a sufficiently adequate representation of the system and therefore Nonlinear Model Predictive Control (NMPC) techniques have to be used. In this article, two techniques of NMPC, namely successive linearization nonlinear model predictive control (SLNMPC) and wiener nonlinear model predictive control (WNMPC) are applied to nonlinear process systems. The major advantage of the two methods being that the NMPC problem is reduced to a linear model predictive control (LMPC) problem at each time step which thereafter allows the optimization problem to be solved using quadratic programming (QP) techniques. Another advantage of these methods is the reduced computational time in calculating the control effort which makes them suitable for online implementation. Both simulation and experimental results show the superiority of the SLNMPC over WNMPC in handling process nonlinearity. The work also shows the favourable performance of the NMPC over LMPC, as expected.