Model predictive control method for continuous nonlinear systems

This thesis develops a non-linear model predictive control method using Taylor series approximations. This method is similar to previously published methods but includes control effort and terminal control conditions in the penalty function. The method results in an analytic approximate solution to the optimal control. A mass-damper example simulated in Matlab is included to demonstrate these findings. It is shown that addition of control penalties provides a more intuitive method for shaping control when compared to extension of the prediction horizon required in previous methods. In addition, accuracy of the approximate solution is improved by directly penalizing control and maintaining a short prediction horizon.