The Research On Tracking Algorithms For Single Aerial Maneuvering Target

Aimed at the problem of the single aerial maneuvering target tracking, this paper starts with the models of maneuvering target, and proceeds from the linear filtering methods to the nonlinear filtering methods, from the single model tracking algorithms to the multiple model tracking methods gradually in Bayesian estimation framework. The following research work has been finished:1. Referring to the idea of modeling of maneuvering frequency, this paper proposes an improved cutting-off normal probability density model (ICN model), and compares this model and the commonly used current statistical model through Monte Carlo simulation. The research result shows better tracking accuracy of the tracking algorithm based on improved cutting-off normal probability density model (ICN-KF algorithm).2. Two tracking algorithms are put forward under the linear filters based on the ICN model. The first kind is the Kalman filtering algorithm based on fuzzy neural network of information fusion (FAnn-KF). The second kind is S(k) multi-scale Kalman filtering (S(k)-MKF) algorithm. Through Monte Carlo simulation with respect to the basic Kalman filtering algorithm, the effectiveness and superiority of the two algorithms are proved.3. In nonlinear filtering condition, the S-amended unscented Kalman filtering (SUKF) algorithm is put forward, which is based on the basic unscented Kalman filter, referring to S-amended thought of Kalman filtering in linear filter. And the research on this algorithm in contrast to the other two kinds of nonlinear filtering algorithm of UKF and Sigma point particle filtering (SPPF) is carried out through Monte Carlo simulation, which verifies the validity and superiority of this algorithm.4. Two kinds of variable structure multiple model algorithms are designed, adopting the model set adaptive strategy of adaptive grid, based on two cases observation of the Cartesian coordinates and polar coordinates, and the above filtering algorithms. The first one is the algorithm of adaptive grid and fuzzy interacting multiple model based on S-amended Kalman filter (AG-FIMM-SKF) under the linear measurement condition. The second one is the algorithm of adaptive grid interacting multiple model algorithm based on S-amended unscented Kalman filtering (SUKF-AGIMM). And Monte Carlo simulations are conducted to verify the effectiveness and superiority of the two algorithms.