Abstract:

Adaptive “current” statistical model algorithm is not the really adaptive target tracking algorithm, the performance of the algorithm depends on the key parameters. In this paper, the maneuvering targets are modeled by the constant acceleration model, and considering the special structure of the process noise covariance matrix, a simplified autocovariance least-squares method is proposed to estimate noise covariances. And this method establishes a relationship between the autocovairnace of the innovation and the unknown covariances, thus, the noise covariance can be estimated by the least-squares method. The simulation results show that, when the maneuvering targets with unit-step acceleration or variable acceleration, the accuracy of the proposed method is better than the adaptive “current” statistical model algorithm.

Key words: Target tracking, Kalman filter, Simplified autocovariance least-squares, Process noise, Noise estimation, Constant acceleration model