Ali Turker Kutay

A Comparison of Two Novel Direct Adaptive Control Methods under Actuator Failure Accommodation

In flight control systems it is crucial to accommodate actuator failures for safe operation. Modern flight control systems are equipped with multiple power sources and actuation systems that offer redundancy in the event of actuator failures. Traditional approaches to flight control system design involve scheduling of fixed gain controllers and control allocation to take full advantage of redundant actuators. Alternative adaptive control approaches have been extensively investigated recently to provide enhanced fault tolerance by augmenting an existing flight control design with adaptive elements. This paper provides a brief background of the state of the art in adaptive fault tolerant flight control and presents a comparison study of several novel direct neural network based adaptive control methods.

Vortex Model Based Adaptive Flight Control Using Synthetic Jets

Presented at the AIAA Guidance Navigation and Control Conference, Chicago, Illinois, August, 2009.

Optimal path tracking control of a quadrotor UAV

This paper presents the linear quadratic tracking (LQT) control of a quadrotor UAV by solving discrete time matrix difference Riccati Equation. First, the nonlinear dynamic model of the quadrotor is obtained by using Newtons equations of motion. Then, the nonlinear dynamic model is linearized around hover condition. The linearized dynamic model is used to solve the optimal control problem. A trade off between good tracking performance and energy consumption is made while defining the performance index (cost function). Time-variant optimal control gains are found off-line by solving discrete time matrix difference Riccati Equation backwards in time. Finally, to validate optimal control system, simulations are performed by using the nonlinear dynamic model as plant and time-variant optimal control gains as state feedback control. The optimal control algorithm used in this paper uses time-variant control gains instead of fixed (time-invariant) control gains used in classical LQR control. Simulations show that, good tracking performance is achieved while decreasing energy consumption compared to the fixed gain LQR control. Some other advantageous properties of the optimal control system used in this paper compared to the fixed gain LQR control are also analyzed. In addition, disturbance rejection properties of the optimal control system are also studied. All algorithms and simulations are done by using MATLAB software.

Aerodynamic Modeling and Parameter Estimation of a Quadrotor Helicopter

This study focuses on aerodynamic modeling of a quadrotor helicopter and the estimation of the model parameters in wind tunnel tests for hover, vertical climb, and forward flight conditions. The motion of a quadrotor is mainly affected by the aerodynamic forces and moments generated by rotors. Accurate calculation of rotor loads is essential for high fidelity simulation of a quadrotor. Momentum and blade element theories are used to obtain expressions for rotor forces and moments for a traveling vehicle. The parameters of the models are then identified through wind tunnel tests where the forces and moments created under various wind conditions and rotor speeds are measured with a six axes balance system.

Closed- Loop Aerodynamic Flow Control of a Free Airfoil

Transitory flow arising from the dynamic response of a free-moving airfoil model to commanded pitch and plunge maneuvers is investigated in wind tunnel experiments. The airfoil is mounted on a 2-DOF traverse and its trim and dynamic characteristics are controlled using position and attitude feedback loops that are actuated by servo motors. Commanded maneuvers are achieved without moving control surfaces using bi-directional changes in the pitching moment over a range of angles of attack that are effected by controllable, nominally-symmetric trapped vorticity concentrations on both the suction and pressure surfaces near the trailing edge. Actuation is applied on both surfaces by hybrid actuators that are each comprised of a miniature [O(0.01c)] obstruction integrated with a synthetic jet actuator to manipulate and regulate the vorticity concentrations. The present work focuses on the transitory response of the flow to step-modulated changes in the actuation input while the model’s position is maintained using the systems controller. Flow control effectiveness is demonstrated by the closed-loop response in plunge to a momentary force disturbance which is analogous to the free flight response to a sudden gust.

Simulation of Rapidly Maneuvering Airfoils with Synthetic Jet Actuators

Synthetic jet actuators are investigated for rapidly maneuvering airfoils that are regulated by a closed-loop control system. To support active flow-control simulations performed here, the closed-loop system and vehicle dynamics are coupled with computational fluid dynamics. High-frequency sinusoidal pitching simulations with and without synthetic jet actuation indicate that the current synthetic jet actuators provide bidirectional change in aerodynamic forces during rapid maneuvers whose time scales are of the same order as the flow time scales. Responses of a wind-tunnel airfoil are well represented in the current simulations, which allows us to predict the response of the system for dynamic conditions representative of free flight. The control system is able to execute rapid free-flight maneuvers. It is observed that the controller is responding to small fluctuations caused by vortex shedding from the trailing-edge actuators.

Modeling and Simulation of a Quadrotor using Curve Fitting Method

This study focuses on aerodynamic modeling of a rotor for an AscTec Hummingbird quadrotor used for controls research at the METU Aerospace Engineering Department through the wind tunnel tests and to develop a high fidelity simulation model for the quadrotor. The main factors which determine the movement of a quadrotor are the aerodynamic forces and moments in three axes created by four rotors. Hence, accurate calculation of rotor forces and moments in varying flight conditions are essential to get a precise simulation of the quadrotor. Instead of using analytical models for aerodynamic forces and moments of a rotor derived using blade element and momentum theories, a second order polynomial rotor model has been developed by MATLAB Curve Fitting algorithm. The developed empirical model allows us to accurately predict aerodynamic loads on a quadrotor in various rotor speeds and flight conditions. Thanks to the new method used in this study, the parameters are integrated easiliy into the simulation of the quadrotor in different flight conditions. Besides, an accurate simulation is conducted by taking into account hub force, rolling moment, and body effect which are usually ignored.

Robust model following control design for missile roll autopilot

This paper presents a robust model following control method augmented with error integration and Luenberger observer for anti-air missile roll autopilot designed using optimal control laws. The design is shown to be robust to external disturbance, noisy measurements and sensor lags by frequency domain analysis. The regulation performance of the controller is presented by simulations.

Adaptive control algorithm for linear systems with matched and unmatched uncertainties

In this paper, a new uncertainty identification method is introduced for both matched and unmatched uncertainties in an uncertain dynamical system. Online identifications of matched and unmatched uncertainties that can be linearly parameterized are ensured without requiring persistent excitation (PE) condition. Furthermore, constant weight matrices that parameterizes the unstructured uncertainties are guaranteed to stay bounded without PE. Findings are implemented on a hybrid adaptive control design, and global exponential stability is established.

Guaranteed exponential convergence without persistent excitation in adaptive control

In this paper, a new adaptive control framework for linear systems in which the matched uncertainty can be linearly parameterized is introduced to guarantee the global exponential stability of reference tracking error and parameter convergence error without requiring restrictive persistent excitation condition. The framework uses time histories of control input and system signals to construct least-squares problem based on recorded data. Then, unique solution to least-squares problem is computed, and assigned as pre-selected value in well-known σ-modification term. Such indirect use of recorded data matrices results in globally exponential convergence of tracking error and parameter convergence error provided that the recorded matrix satisfies the simple rank condition. The proofs are given by Lyapunov stability theorem, and the results are illustrated with simulations.