Ballistic target tracking method based on variable structure adaptive multi-model box particle filter

Abstract

The invention provides a ballistic target tracking method based on a variable structure adaptive multi-model box particle filter. The method comprises the following steps of: S1, calculating a transition probability between model sets; S2, obtaining a variable structure model set; and S3, carrying out an interactive multi-model box particle filtering algorithm. The invention has the advantages that a variable structure adaptive multi-model box example filtering (VSAIMM-BPF) tracking method is applied, full-stage continuous and continuous tracking of a trajectory target is achieved, and the method obtains high tracking precision while reducing calculation complexity. The invention is more in line with the actual state of the current system in the distributed multi-sensor network, tracking errors generated during model switching are effectively reduced, and compared with a traditional tracking algorithm, the invention can stably and continuously track the whole flight stage of the ballistic target. Especially in the aspect of calculation efficiency, the efficiency of the VSAIMM-BPF algorithm is improved by nearly three times under the condition that similar precision is achieved, and the method is particularly suitable for a multi-sensor task planning system with a high real-time requirement.

Description

technical field [0001] The invention belongs to the technical field of target tracking methods, and in particular relates to a ballistic target tracking method based on a variable structure self-adaptive multi-model box particle filter. Background technique [0002] In the current anti-missile combat, there is a nonlinear relationship between the sensor measurement value (including radial distance, pitch angle, azimuth angle) and the target state, and the tracking system often has unknown synchronization deviation in the complex confrontational battlefield environment Or in the case of system delays, the state noise and observation noise are also non-Gaussian. The above problems belong to nonlinear filtering problems. Currently, commonly used filtering algorithms include Extended Kalman Filter (EKF) algorithm, Unscented Kalman Filter (Unscented Kalman Filter, UKF) algorithm and related improved algorithm (Kalman filter and The point estimation algorithm represented by the o...

AI Technical Summary

Problems solved by technology

Studies have shown that EKF has good real-time tracking performance, but has poor adaptability to nonlinearity and is prone to divergence; UKF has better tracking effect, but it is also more complicated for high-dimensional situations; PF can achieve high tracking accuracy, but there is a large amount of calculation , poor real-time performance and particle degradation

Images