Improved Adaptive Estimation Approach for Aircraft and Land Vehicle Applications

Abstract

A precise dynamic model and exact statistical information are required in a standard Kalman filter to ensure optimal performance. Without these, degraded performance may be obtained. In this paper, an improved adaptive estimation approach for uncertain models is proposed, and the fading matrix is adaptively estimated according to the observability of the system state. Meanwhile, a sequential way to calculate the covariance matrix of the measurement noises is performed, thereby strengthening its ability to resist disturbances further. The proposed algorithm was implemented in practical data processing of GNSS (Global Navigation Satellite System) and INS (Inertial Navigation System) integrated systems. The measurement sensors were equipped on aircraft and land vehicles to test their stability and robustness. Comparisons with existing fading filters and conventional nonlinear Kalman filters show that the proposed algorithm controls model errors, weakens influences of constantly changing and abrupt outliers, achieves relatively high precision, and restrains filter divergence effectively.

Keywords

• Adaptive estimation
• nonlinear Kalman filter
• fading matrix
• GNSS/INS
• integrated navigation and positioning