PECASE: Robust Intelligent Control, Demonstrated for Reconfigurable Flight Systems

Duke University, Durham NC

Investigators

Abstract

Recent advances in a variety of technologies and applications call for improved performance and reliability, while exacerbating the complexity and uncertainty of systems and their surroundings. Although classical control theory already allows for a high degree of machine automation, a renewed interest in systems that display intelligent behavior is emerging across disciplines. The proposed research will develop a novel adaptive control approach that approximates the nonlinear control law by a sigmoidal neural network, and optimizes its performance on line, subject to actual plant dynamics. By combining approximate dynamic programming and integral quadratic constraints, it can be shown that the neural network controller is characterized by the same closed-loop stability and robustness guarantees as classical H-infinity controllers. Ensuring that these properties are preserved during adaptation will make the adaptive design viable for applications where safety is a leading concern, such as, reconfigurable aircraft control. As part of the proposed research, the adaptive control approach will be demonstrated not only through numerical simulations, but also through experiments. The first testbed will consist of a cantilevered beam with piezoelectric devices used as disturbance and control sources. In the experiments, the effectiveness of the adaptive controller for reducing the vibration of the first two natural modes will be tested and compared to that of controllers with parameters held fixed. After considering this low-dimensional system, the approach will be implemented to design an adaptive flight controller based on the model of an experimental fly-by-wire Raytheon Bonanza aircraft. The flight controller will be tested on a piloted simulator developed by an experienced research group at Wichita State University (WSU). Subsequently, the PI will work with the WSU group and Raytheon Aircraft Company on interfacing the control software with the on-board flight computer, and on testing it on the Bonanza through a set of experimental flights. Research and education on control theory and practice will be integrated by designing a set of experiments for undergraduate and TIP high-school students that employ the following testbeds: a vibrating cantilevered beam, an acoustic enclosure, and a flapping micro-aerial vehicle. The broader impact of this project is the creation of initiatives that expose students from K12 through graduate school to advanced control research and experiments being conducted in the aerospace industry and at NASA. By leading undergraduates and TIP students through these experiments, the PI's graduate students will become important role models for a diverse and gifted group of youngsters.