Tracking of optimal trajectories for power plants based on physical models

The paper deals with tracking of optimal trajectories for large scale non linear systems like power plants. It is assumed that an accurate nonlinear model of the plant is available, but because of its size and complexity, it cannot be used directly for long term dynamic optimization. For this reason, reference input and output trajectories are obtained from a simplified optimization model. Then a tracking Model Predictive Control (MPC) algorithm based on tangent linear approximations of the nonlinear model along a nominal trajectory is used to correct the trajectories. It includes input and output constraints, state estimation and disturbances rejection. The concept is shown on the tracking of optimal trajectories by a Combined Heat and Power (CHP) plant with heat storage and time varying electricity price.