THE EFFECT OF TERMINAL WEIGHT ON THE PREDICTION HORIZON OF A GAS TURBINE ENGINE USING MODEL PREDICTIVE CONTROL

摘要

Robust Stability Margins of a Model Predictive Control (MPC) based gas turbine engine control system can be guaranteed as long as the prediction horizon is infinite. Unfortunately, the infinite prediction horizon is impossible to implement in the real world. Even a limited prediation horizon can require computation times that make real time embedded MFC based gas turbine applications infeasible to implement. An ideal situation would be to reduce the prediction horizon so that sufficient stability robustness is realized, computation time is reduced and, most importantly, the system exhibits good dynamic response characteristics. This paper will examine if the selection of a Terminal Weight is able to reduce the prediction horizon (and thereby reduce computation time) while maintaining sufficient stability robustness and dynamic response. The selection of the terminal weight will be based upon the Linear Quadratic Regular (LQR) approach. A simplified model of a typical (generic) gas turbine engine will be used as the plant. The model is "sanitized" so that the model outputs are presented in terms of normalized (or scaled) parameters.