Synthesis of an Intelligent Algorithm for Estimating the Angular Orientation of a Multirotor UAV with an Adaptive Dynamic Model Based on a Neural Network-Identifier

The growing popularity of multirotor unmanned aerial vehicles (UAVs) can be explained by their versatility and high maneuverability compared to other robotic platforms. High requirements for the technical characteristics of the UAV require the development of new highly efficient algorithms to use all onboard capabilities. One of these requirements is the high accuracy of the orientation system, which is one of the main elements of the avionics in any UAV. It is known that the adaptive motion model makes it possible to increase the accuracy of the estimation in comparison with the classical Kalman filter. Also, the quality of the filter functioning is determined by the intensity of external influences in the dynamics model. This determines the need for multiparametric adaptation of the evaluation algorithm by changing the magnitude of the dissipative component and the intensity of external influence. The aim of the work is to synthesize an intelligent algorithm for estimating the parameters of multirotor UAVs with model adaptation based on a multiparametric neural network-identifier.