Wind Disturbance Estimation and Compensation for Quadcopters using Fuzzy Neural Network

In this paper, the fuzzy neural network (FNN) with reinforced rules is used to estimate the wind disturbance of quadcopters in real time, and the wind disturbance compensation is performed on the position loop, attitude loop, and lift loop according to the estimated disturbance. Among them, the neural network part is used for data-driven modeling of complex aerodynamic effects of quadcopters, the fuzzy part is used to improve the interpretability of the conventional neural network, and the reinforced rules are used for improving the generalization ability of conventional FNN to quadcopters whose computational resources are limited. Simulation and comparative studies have been conducted to verify the effectiveness and merits of the proposed method.