Designing an AI-based Fractional Controller for an Uncertain MIMO Nonlinear System

A quadrotor is a robotic vehicle with four rotors capable of vertical landing and take-off. The four rotors are infour corners, rotating clockwise and counterclockwise. Quadrotors are mainly considered as benchmarks forevaluating designed controlling algorithms. Couples of rotors are rotating in opposite directions to balance the torque.By controlling each rotor's speed, different movements can be made, and the amount of vertical propulsion can beadjusted. Quadrotor has non-linear coupled dynamics, and uncertainties are an inseparable part of the vehicle'sbehavior. These complexities lead to the implementation of the various linear and non-linear methods for controllingthe device. In this paper, according to the equations related to the dynamics of the quadrotor, we will examine themotion and position of the quadrotor using three types of controllers. The success of these controllers depends on theproper selection of their gains. In this article, we have used the neural networks to find the optimal gains, which receivethe error and its changes as input and change them in an online mode. Comparison of simulation results withconventional PID and FOPID controllers indicates the superior performance of the intelligent FOPID controller inincreasing the stability of the quadrotor robotic system.

Keywords:

Robotic System , FOPID , Neural Networks