Global probability data correlation method used for passive multi-sensor target tracking

Abstract

The invention discloses a global probability data correlation method used for passive multi-sensor target tracking. By using the method, the error caused by local linearity of a nonlinear part of the original method and incomplete probability distribution in data association due to measuring incompletion observability can be effectively overcome; and the unreasonable arrangement of the tracking gate caused by the biased probability distribution estimation in the original method is also revised in the method. In addition, the invention also provides a multi-sensor sequential processing sequencing mechanism, which can effectively improve the target tracking effect. The method is especially suitable for passive wireless sensor network target tracking system with low original data measuring reliabilities.

Description

technical field [0001] The present invention relates to a global probability data association method for passive multi-sensor target tracking, in particular to a global probability data association based on target position prediction suitable for passive wireless sensor network target tracking systems with low reliability of measurement raw data method. Background technique [0002] Target tracking theory and technology have important application value in both civilian and military fields, and this field has always been the focus of development in academia, industry and even the military. Traditional active target tracking systems, such as radar, obtain target azimuth and distance measurements by transmitting radar waves and receiving reflected echoes. The position of the tracking system is easily detected by the enemy and then destroyed by the enemy. . Therefore, in this application scenario, passive sensors embody great advantages. However, the passive sensor can only o...

AI Technical Summary

Problems solved by technology

[0018] Although the existing probabilistic data association method for passive sensors can realize the target tracking function, there are some problems in directly using the traditional probabilistic data association method due to the defect that the measurement of the passive sensor is not completely observable to the target state: first, The corresponding relationship between the measurement of the passive sensor and the target state is nonlinear, and local linearization is only an approximate approximation, so the measurement residual covariance There is a certain error in the calculation of the passive sensor; secondly, because the measurement of the passive sensor is only a 1D signal, part of the information is missing compared to the 2D target position, and the distribution of the measurement residual covariance is generally not related to the predicted measurement direction angle. Symmetry, this condition is implicitly assumed in the existing methods, which causes a certain deviation in the setting of the tracking gate and the calculation of the correlation probability; in addition, for the application of multi-sensors, the sequential processing method of the sequential processing structure needs to be processed in a reasonable order. Each sensor measurement is processed. If the processing order of the sensor measurement is randomly determined, the influence of the processing order on the result in the sequential processing method is ignored. Obviously, the effect is not ideal. The incompleteness of passive sensor measurements, such incomplete observations may lead to a small contribution to the update of the target state, so the ranking method is not globally optimal

Images