Model Predictive Control of a Nonlinear Aeroelastic System using Reduced-Order Volterra Models

摘要

This paper investigates the implementation of model predictive control strategies in order to control the pitch response of a nonlinear aeroelastic system. Model predictive control, also known as receding horizon control, entails computing an optimal sequence of control inputs over a finite time horizon in order to minimize a cost functional subject to constraints. This approach requires a dynamic model for the system in order to predict the system response to candidate control input sequences. In this paper, a simulated nonlinear aeroelastic system is modeled in terms of third-order Volterra series, and these Volterra models are implemented in a model predictive control algorithm to control the pitch response of the system. The Volterra model is obtained from simulated input - output data using a multiresolution-based kernel identification algorithm that has previously been developed by the author and used to generate Volterra models from both simulated and flight test data from several aeroelastic systems. This paper investigates the feasibility of using Volterra-based model predictive control strategies to control a simulated nonlinear aeroelastic system corresponding to a model of the Texas A&M Nonlinear Aeroelastic Testbed. The model predictive control algorithm is used to compute trailing edge flap deflection commands in order to control the pitch response of the system at different free stream velocities, including a flight condition for which the underlying linear aeroelastic system is in an unstable flutter condition.