Metodologia para modelagem computacional de Kalman Fuzzy baseada em triangulação de fontes de informação para rastreamento e previsão de comportamento em sistemas suborbitais

Abstract

Suborbital rockets are launched carrying a payload to the microgravity environment. In addition to the applications of aerospace activity and access to space, the rocket launch activity promotes scientific and technological development. Vehicle tracking data is monitored by Flight Security, primarily used to determine whether the flight is within the prescribed safety limit or whether there is a need to terminate the flight by destroying the vehicle. Currently, the Alcantara Launch Center (CLA) performs the tracking of aerospace vehicles through systems with tracking radars. In order to increase the level of availability of vehicle location data, this work proposes an approach for determining the position of the rocket-based on antenna triangulation. Therefore, the rocket’s position is estimated based only on the azimuth and elevation data of the antennas pointing in its direction. The result of triangulations is input signals for a dynamic filter that aims to achieve accuracy close to the radar signal. This dynamic filter must consider inherent nonlinearities to the trajectory dynamics. To handle such nonlinearities, a fuzzy clustering technique was used to partition the trajectory and then use linear filters for each partition and a fuzzy system to manage such clusters. System identification techniques have been applied to convert a signal from the triangulation step into a radar-like signal and reduce errors and noise. Simulations were carried out to demonstrate the performance of the proposed method using historical actual flight data from a suborbital vehicle. The implementation demonstrates that the technique estimates the vehicle position based only on azimuth and elevation data from antennas pointing to the vehicle. Also, the dynamic filter design reduces errors and noise in rocket estimation.