NONLINEAR INTO STATE AND INPUT DEPENDENT FORM MODEL DECOMPOSITION: Applications to Discrete-time Model Predictive Control with Successive Time-varying Linearization along Predicted Trajectories

摘要

Linearization techniques are well known tools that can transform nonlinear models into linear models. In the paper we employ a successive model linearization along predicted state and input trajectories resulting in linear time-varying model. The nonlinear behaviour is represented in each time sample by recurrent set of linear time-varying models. Solution of the optimal non-linear model predictive control problem is obtained in an iterative way where the most important step is the linearization along predicted trajectory. The main aim of this paper is to analyse how the nonlinear system should be transformed into linear one to ensure possibly fast solution of the model predictive control problem based on the successive linearization method.