Identification of Autonomous Air Refueling Using Parametric Sigmoid Neural Networks Considering Turbulence Effects

Abstract

In this research, the problem of improving probe and drogue identification in aerial refueling systems in the presence of uncertainty is investigated. This uncertainty is caused by the effects of environmental turbulence and tanker aircraft turbulence on the aerodynamic parameters of the refueling aircraft. The purpose of this work is to identify aerial refueling aircraft and extract control parameters for the development of the automatic aerial refueling system of tanker aircraft. To do this, a non-linear model of receiver airflow is developed, taking into account tanker airflow and environmental turbulence effects. In the second step, the desired algorithm is designed by combining a nonlinear neural network with a parametric sigmoid function. Automation of such a scenario requires many activities to be performed by algorithms that are delegated to trained crews. After mathematical modeling and obtaining the input and output values with MATLAB software, learning algorithms have been used to train them with the network. In this case, after each repetition, the average error of the network in the output production is reduced to achieve convergence. The training error is usually reduced in the initial stage of the training connection. By comparing the results, it can be seen that the proposed method has a better performance than conventional methods in reducing the mode errors between two planes and limits all signals uniformly.