An interactive multi-model probability data association based on Kullback-Leibler differential entropy for reliable positioning is proposed to stably track noise. When the changes in measurement noise and process noise occurs, the interactive multi-models can fuse the estimated results and by combination of probabilities in models, noise can be dynamically tracked, to improve the reliability and accuracy of positioning procedures. The simulation results show that with the increase of measurement noise, the KL-CIMM-PDA method is more stable and more adaptive to track noise than the CIMM-PDA method and compared with the IMM-UKF, the accuracy of the KL-CIMM-PDA is improved by 26.7%.

为了稳定地跟踪噪声，提出了一种基于Kullback-Leibler微分熵的交互式多模型概率数据关联方法。当测量噪声和过程噪声发生变化时，交互式多模型可以融合估计的结果，并且通过组合模型中的概率，可以动态跟踪噪声，从而提高定位过程的可靠性和准确性。仿真结果表明，随着测量噪声的增加，与CIMM-PDA方法相比，KL-CIMM-PDA方法更加稳定，对跟踪噪声的适应性也更高，与IMM-UKF相比，KL-CIMM-PDA的准确性更高。 PDA改善了26.7％。