Longitudinal Aerodynamic Parameters Estimation using Machine Learning with Neuro Artificial Bee Colony Fusion Algorithm (NABC)

This paper provides a novel estimating approach for longitudinal aerodynamic parameters in the presence of system and measurement uncertainty. The proposed method dubbed as NABC is a hybrid technique that employs biologically inspired optimization, namely Artificial Bee Colony (ABC) in fusion with widely used Artificial Neural Network. For parameter estimation, the motion variables and control inputs are initially used to train the neural network, and the corresponding predicted output of aerodynamic coefficients is obtained. The method is first validated on the simulated dataset intentionally contaminated with 5% and 10% of random noise, resembling the actual flight data. Then, the algorithm's efficacy is tested on the real flight data of ATTAS aircraft. The aerodynamic parameter estimations obtained using the proposed method are compared to the benchmark estimate methodologies, like Least Squares and Maximum Likelihood Method (MLE). The confidence in the estimates generated using the suggested technique has been strengthened by statistical analysis of the parameter estimations.