Modeling and identification of highly maneuverable fighter aircraft dynamics using block-oriented nonlinear models

Abstract

In this paper, a new approach based on block-oriented nonlinear models for the modeling and identification of aircraft nonlinear dynamics is proposed. Some of the block-oriented nonlinear models are regarded as flexible structures, which are suitable for the identification of widely applicable dynamic systems. These models are able to approximate a wide range of system dynamics. In general, aircraft flight dynamics is considered as a nonlinear and coupled system whose dynamics—in addition to pilot control inputs—depend on the flight conditions such as Mach number and altitude, which cause the aircraft dynamics to have various operational points. In this study, three types of block-oriented models, namely the Hammerstein, Wiener, and Hammerstein–Wiener models with different nonlinear functions, have been used and compared in order to identify and model the aircraft nonlinear dynamics. These models have been employed in three forms of single-input single-output, multi-input multi-output, and multi-input single-output (MISO); of which, multi-input single-output has been recognized to have fewer errors in aircraft nonlinear dynamics identification. Thus, it has been demonstrated that six separate multi-input single-output models (with three inputs and one output), which have been trained with experimental flight test data, can model the coupled nonlinear six-degree-of-freedom dynamics of a highly maneuverable aircraft.

A self-organizing multi-model ensemble for identification of nonlinear time-varying dynamics of aerial vehicles

“…19 More specifically, the control and stability derivatives of an aircraft and the dynamic pressure change by altering these two scheduling variables, bearing in mind that a dynamic model obtained at a specific operating point is not valid in the entire flight envelope. 37 In such conditions, it is possible to calculate the system output at an arbitrary operating point (in the intended flight envelope) using an MME, where each local model of the MME corresponds to a specific operating point. The selected operating points for the local models should be such that they can cover the entire flight envelope.…”

Identification of nonlinear time‐varying systems using wavelet neural networks

“…More precisely, the stability and control derivatives of an aircraft dynamic model vary by changing the flight speed. Further, the air density changes by altering the aircraft altitude, which in turn, leads to changes in the dynamic pressure 2 . Thus, an aircraft has a nonlinear time‐varying dynamic model, and an obtained dynamic model in a specific operating point is not valid in the entire flight envelope.…”