AIRCRAFT FLIGHT PATH PREDICTION UNDER COMPLETE PARAMETRIC UNCERTAINTY

Abstract

Most of the methods for predicting the behavior of dynamic systems are based on the information
about the parameters of their mathematical models. However, the problems of
nonstationarity, nonlinearity and nonidentifiability of models of real complex systems lead to the
fact that traditional parametric methods are applicable in practice only when the parameters and
structure of models of systems are reliably known, and the uncertainties in the formulation of the
problem are significantly limited. The article describes an original nonparametric method for
predicting the aircraft flight path under absence of a priori information about the parameters of its
mathematical flight dynamics model. The proposed method, unlike similar widely known ones,
does not use logical or statistical calculations and does not require its preliminary training or
long-term tuning. It is based only on the basis of a retrospective analysis of several sequential
values of the spatial coordinates of the aircraft and its control signals, therefore it is not subject to
model errors and can be used to predict the flight path of the aircraft under complete parametric
uncertainty, even in the case of non-identifiability of its flight dynamics model. The results of numerical
simulation of the solution to the problem of predicting the flight path of an unmanned
aerial vehicle of the most common type of quadrocopter under complete uncertainty in parameters
of its mathematical model are presented. The results obtained confirm the efficiency of the developed
method and show high performances of the accuracy of solving the problem and the speed of
tuning the algorithm. The described approach can be used to predict the motion path of any other
vehicle (car, ship, etc.), if its model is linearizable over the observed time interval and there is
information about its control signals. Practical implementation of the described nonparametric
method together with traditional parametric ones will improve the accuracy of flight path predicting
and solve the problem of high-precision landing of an unmanned aerial vehicle on an actively
maneuvering ship, and specifically in the event of various critical situations.