Analyzing fault tolerances by developing linear and nonlinear controllers for multirotor systems

The usage area of Unmanned Aerial Vehicles (UAV) is increasing day by day. Quadrotors in particular are used for a wide variety of applications, including security, traffic surveillance, natural risk management, environmental exploration, agriculture and military. However, engine malfunctions, sensor malfunctions or communication malfunctions may occur in quadrotors, causing serious accidents in vehicles and damage to the vehicle and to the environment. The aim of this study is to develop a fault- or noise-tolerant controller design for a quadrotor to increase its reliability and stability in case of failure or disruptive external factors. In order to derivate the state space equations of the quadrotor, the equations of motion were calculated based on the x,y,z coordinate system in free space. In these movements, the mathematical equations of the roll, pitch, yaw and altitude movements of the four-rotor system are defined. The defined equations were designed in Matlab Simulink and system simulations were made. Linear and non-linear controllers have been developed to control the system. First of all, a traditional PID feedback control system was implemented. In the analysis with the gain coefficients we determined, the controller has the ability to correct errors to a certain extent under restrictions and able to follow the input reference signal. Gain coefficients were determined again with the Simulink PID block, and the outputs of the system's roll, pitch, yaw and movements were examined in response to different reference signals. Although better results were obtained in the analysis made with the PID block, it was observed that the system was more stable at the gain values We determined for altitude movement. With non-linear controller methods, better results can be obtained in terms of flight safety in complex systems with many dynamics such as quadrotors. For this reason, in the thesis study, a feedback controller was designed for altitude movement and the results were compared with the results of the traditional PID controller system. It has been observed that the system with a nonlinear controller provides better results in terms of tracking the reference signal and controlling the stability of the system against disturbing effects. When we adjust the PID controller values, we can obtain robust state responses, but the reference tracking of the system is prolonged. By giving the input signal to the system in different forms, the movement of the quadrotor in free space was observed and interpreted. Unlike other studies, the system, which has multiple non-linear inputs and multiple outputs, is considered as a linear system with the input-output measurement method and the altitude movement transfer function is derived. A Kalman Filter was designed from this function and the results were examined. As a matter of fact, with the designed filter, the state estimation of the system was made only with the signals received from the input and output of the system, and more noiseless signal was received at the system output. As a result, more stable outputs were obtained with backstepping control, one of the nonlinear control methods, compared to linear control methods, and the system state was estimated using a filter and the effect of system disturbances was minimized.