Aiming at the problem that ultra wide band（ UWB） positioning technology is highly impacted by nonline of sight（ NLOS） factors from an environment

which results in low positioning precision and poor positioning stability

an approach is proposed for accurate dynamic target positioning by applying particle filtering twice on the raw data collected by the UWB. The first particle filtering is used to save the relational model between the measured distance and the error model of each anchor. Based on the model

the second particle filter is used by assigning particles with different weights to improve the positioning accuracy. Experimental results in different environments show that the proposed algorithm can effectively suppress the influence of NLOS error. With this algorithm implemented

the positioning accuracy of the dynamic object reaches 0. 29 m at the moment when NLOS errors appear to be serious

which is an improvement of 48. 21 % and 17. 14 % when compared with the triangulation algorithm provided by the UWB system and traditional particle filtering-based approach. The dynamic object positioning accuracy is remarkably improved in NLOS environment.