Improvement of hovering stability for UAVs under crosswinds via evolutionary learning-based optimal PID control

Abstract

This study aimed to optimize PID gain values to enable the swift recovery of unstable unmanned air vehicles (UAVs) while maintaining robust control in the presence of crosswind effects. Previous research concentrated on rotor and blade optimization for thrust. In this study, PID gain values were obtained by generating a PID control algorithm for the optimized UAV. A surrogate model between PID gain levels and performance variables was constructed by using a backpropagation neural network. Moreover, a nondominated sorting genetic algorithm-II was used to optimize PID gain settings for stability in a UAV affected by crosswind and instability. Altitude disturbance during hovering increased by 78 %, with roll motion and crosswind effects rising by 19 % and 33 %, respectively, in comparison with initial designs. Evolutionary learning-based PID control improved UAV stability, enabling quick recovery from crosswind-induced tilting. In future research, PID control gains with high robustness against external disturbances for various flight conditions will be determined.