PSD Estimation Based Enhanced Kalman Filter for Target Tracking With Singer Acceleration Noise

초록

This study proposes a new adaptive Kalman filter for high-precision cooperative target tracking based on the Singer acceleration model. In cooperative scenarios, the maneuver time constant alpha is known in advance through mission planning or communication between targets, allowing a physically consistent and structure-preserving noise modeling. Traditional approaches often necessitate the estimation of the full noise covariance matrix, which can be computationally intensive and prone to inaccuracies and spurious state correlations. To overcome this limitation, this work elaborates a novel online power spectral density (PSD) estimation scheme. Using known alpha to reduce the number of unknown variables, the new approach can improve the accuracy and reliability of the estimation. Numerical experiments in cooperative target tracking demonstrate that the refined algorithm achieves robust adaptability to dynamic noise that varies over time, provides high-precision state estimates, and maintains low computational complexity.

키워드