One of the malfunctions that can happen in the operating quadrotor is the actuator fault. The fault signal appears in the form of motor power degradation or complete loss. It can have catastrophic consequences, such as a crash, for the operating quadrotor. So, fault signal detection and fault-tolerant control are needed when a faulty situation occurs. In this regard, it is necessary to find out which actuator has failed first.
This thesis describes how to detect fault signals in actuators based on the dynamic model of quadrotors. There are two dynamic models of quadrotors used in this thesis. One is a model that assumes the quadrotor dynamics in an ideal environment. It represents the quadrotor dynamics relatively straightforward, so it is commonly used in most studies using quadrotors. The other is a model that considers the effect of aerodynamic properties generated by the rotation of propellers. Linear state equations are obtained for each two quadrotor models when the quadrotor hovers to apply the fault detection method.
The fault detection method used in this thesis is called a geometric approach, using the subspaces of the error system expressed as the estimation error of the linear system. We use the characteristics of the subspaces of the error system to design suitable filters for actuator fault detection.
To analyze the performance of the designed fault detection filter, a simulation based on MATLAB Simulink was used. We verified the performance of the designed fault detection filters and checked the effects of aerodynamic properties by the rotation of propellers on fault detection performance.
3 Quadrotor dynamics 9 3.1 Classical quadrotor model 10 3.2 Aerodynamic effect of propellers 11 3.3 Modeling of the drag force and drag torque 14 3.4 Linearization and controller design 16
4 Fault Detection Problems 19 4.1 Fault detection problems 20 4.1.1 Algorithm for (C,A)-invariant subspaces 22 4.1.2 Algorithm for unobservable subspaces 22 4.1.3 Solvability condition 23 4.1.4 Design condition of the observer gain G_i 24 4.1.5 Design condition of the filter matrix H_i 24 4.2 Special case when C is full rank 25 4.3 Designing fault detection filters in classical quadrotor model 25 4.4 Designing fault detection filters in detailed quadrotor model 35
Research Interests
Resilient control systems; Control systems with nonlinear sensors and actuators; Quasi-linear control systems; Intelligent transportation systems; Networked control systems