재귀적 주성분분석을 사용한 발사체 추적 필터 설계

초록

In this paper, we propose a launch vehicle position data estimation filter using recursive principal component analysis. Position measurement data collected from tracking systems are filtered to remove abnormal outliers. Conventional filtering methods require prior identification of system and measurement model. However, variations between the models and the actual dynamics of the launch vehicle can lead to increased prediction errors. The proposed method eliminates the need for pre-defined models by generating predictions based on the principal components of the measurement data. To evaluate the performance of the proposed method, the mean residual is compared with that of conventional filters