On the Development of a Fuzzy Logic Model-less Aircraft Controller

摘要

The Robust Learning Control program at NASA Langley Research Center(LaRC) is developing techniques to enable Rapid Aerodynamic Modeling for NASA's Transformational Tools and Technologies Project. The challenge is to design a robust flight control system without prior complete knowledge of the aircraft mathematical model and aerodynamics. In this paper, the challenge is addressed with the development of a Takagi-Sugeno-Kang fuzzy logic controller. This paper compares the performance and robustness of similarly configured Type-1 and Interval Type-2 fuzzy logic controllers. These fuzzy logic controllers are not based upon a mathematical model but rather on a rule base of generic flight control laws generated from the designer's knowledge of aircraft flight mechanics. The controllers must provide acceptable performance without the need for tuning. A systematic approach for selecting input and output scaling gains for the fuzzy inference systems that makes this possible is discussed. The controller architecture uses two channels to provide absolute and incremental controller commands as needed. The absolute channel is designed to reject disturbances and decrease rise time, while the incremental channel provides tracking and reduced steady state error. The fuzzy logic controllers were implemented on an aircraft model in the NASA Langley 12-Foot low speed tunnel mounted on a free-to-pitch and free-to-roll rig. The development of the controller architecture and wind tunnel results will be presented.