Fault-Tolerant Quaternion Neural Network Fast Terminal Sliding Mode Control for Guidance and Navigation of UAV’s under Actuator Faults

This research presents a terminal sliding mode controller and quaternion neural network observer for unmanned aerial vehicle (UAV) guidance and navigation under actuator fault. Initially, to determine which of a quadrotor's four actuators is failing, a quaternion neural network observer is created. The quadrotor dynamic model includes the DC motor dynamics, which is taken into account when designing the fault-tolerant observer. The quaternion neural network presents four failure modes pertaining to an actuator state in which instability and divergence from the healthy state occur when one of the actuators fails. Next, a fault-tolerant terminal sliding mode controller is engineered to deliver the necessary controller action when all actuators are functioning normally, or if one or more quadrotor UAV actuators malfunction. The rapid terminal sliding mode controller is built in order to acquire the diverse control laws when one or many actuators fail or when the system is in healthy state. The suitable Lyapunov functional at various time instants is taken into account throughout the construction of this fault-tolerant controller to restore optimal performance. In addition to the theoretical outcomes of this investigation, a simulation is run to demonstrate the effectiveness of the observer/controller approach and to contrast it with other sliding mode and neural network observer/controller combinations. The research study's conclusions and discussions are given at the end.