A SVD-based extended Kalman filter and applications to aircraft flight state and parameter estimation

Abstract

In this paper, a new robust extended Kalman filtering algorithm based on singular value decomposition (SVD) of a covariance/information matrix is presented with application to the flight state and parameter estimation of aircraft. The presented algorithm not only has a good numerical stability but also can handle correlated measurement noise without any additional transformation. The algorithm is formulated in the form of vector-matrix operations, so it is also useful for parallel computers. The applications to the flight state and parameter estimation by simulated and actual flight test data computation of two types of Chinese aircraft show that the new algorithm presented in this paper can give more accurate estimates of flight state and parameters than an extended Kalman filter (EKF) for different initial values and noise statistics. Moreover, the new algorithm has less requirements for maneuvering shapes, noise levels, data length and better convergency than those of the EKF. The computational requirements for one-step filtering updates of the new filter have been reduced greatly by exploiting some special features of system and measurement models. It is proved that the new filtering algorithm can give good results even for low sample rate flight test data.