Online Identification of Aerodynamic Parameters of Experimental Rockets Based on Unscented Kalman Filtering

Abstract

Online identification of aerodynamic parameters of experimental rockets was completed based on unscented Kalman filtering (UKF). Numerical simulation, hardware-in-the-loop (HIL) simulation, and flight tests were conducted. The identification error of aerodynamic force in numerical simulation and HIL simulation is within 2%. For flight test data, trajectory reconstruction was performed using the identified aerodynamic forces, and the results showed that the identification results were more accurate than the interpolation table calculation results. The flight test identification results show that the identification method can complete parameter online identification under the conditions of limited performance of onboard computers, real sensor errors, and servo response. The approximate linear correlation between $\alpha$ and ${\delta }_e$ and the reason for their formation from the moment balance were analyzed. It was pointed out that when the recognition sampling period is long, this phenomenon will affect the identification of parameters, and a solution is proposed.