Multivariable control of the J-85 turbojet engine for full flight envelope operation

Abstract

The design of a J-85 turbojet engine controller using Edmunds' method and gain scheduling techniques is presented. A nonlinear engine model was first linearized at 50 selected operating points throughout the flight power envelope of the J-85 engine. The operating points were selected based on the engine's altitude, Mach number, and compressor rotor speed. Edmunds' method was then used to design a linear controller for each operating point. The objective of this method is to design a simple controller, which achieves a desired performance criteria. Finally, for full flight envelope operation, two gain scheduling approaches were evaluated: 1) extended Kalman filter and 2) neural networks. The objectives of this research are to evaluate Edmunds' method for the design of a J-85 engine controller, as well as gain scheduling approaches for the full flight power envelope operation. After completing the controller design for 50 operation points, we found that Edmunds' method is very effective for this application. For the gain scheduling approaches, simulation results show that the controller using neural networks has better performance.