The Effect of Using Different Feature Sets on the Fidelity of Deep Learning Based System Identification of a Fighter Aircraft

Accurate aircraft models reflecting flight dynamics are pivotal in various domains such as aircraft design, development, and certification. Achieving high-fidelity results demands the development of flight dynamics models through System Identification using flight data. While Deep Learning-based studies have gained prominence, challenges persist in collecting flight data due to budget constraints and accurately estimating the Non-Linear region. This study aims to enhance flight models in the Non-Linear region and optimize flight test campaigns through the analysis of different feature sets and flight envelopes using Deep Learning methodology.