Multiple Affine Model in Predictive Control of a Nonlinear Dynamic System

In this paper, it is presented a comparative view between Nonlinear Model Predictive Control (NMPC) and Linear Model Predictive Control (LMPC) applied to Multiple Affine Models (MAM) for a nonlinear dynamic system. Therefore, two MPC strategies are employed. In the NMPC scheme, three local optimization methods are used. Two gradient based ones that are Interior Point and Trust Region. The third is the Nelder Mead method which is stochastic. The comparison is realized through simulations on a delayed nonlinear system. It highlights tracking performances attained by each strategy in terms of stability, speed and accuracy. In addition, the comparison concerns the computing times of the optimal control. The LMPC scheme applied to MAM achieves best tracking performances and much lower computing time.