Computer Engineering and Applications ›› 2018, Vol. 54 ›› Issue (4): 44-49.DOI: 10.3778/j.issn.1002-8331.1609-0372

Previous Articles     Next Articles

New distribution interacting multiple model generalized probabilistic data association algorithm

ZHANG Weihua1,2, SUN Qichen1, ZHANG Lijing1   

  1. 1.School of Information & Electrical Engineering, Ludong University, Yantai, Shandong 264025, China
    2.Asset Division, Ludong University, Yantai, Shandong 264025, China
  • Online:2018-02-15 Published:2018-03-07

新的分布交互式多模型广义概率数据关联算法

张维华1,2,孙启臣1,张丽静1   

  1. 1.鲁东大学 信息与电气工程学院,山东 烟台 264025
    2.鲁东大学 资产处,山东 烟台 264025

Abstract: To effectively solve the problems of distributed multi-sensor multi-maneuvering target tracking under the dense clutter scenario, a new distributed interacting multiple model multi-sensor generalized probabilistic data association algorithm based on improved D-S combination rule of evidence(DIMM-MSGPDA-IDS) is proposed. Firstly, a single sensor IMM-GPDA algorithm is used to track multiple maneuvering targets on each local node, and the filtering results of each model, such as the state estimate, covariance estimate, model probability, combination of residuals and the corresponding covariance matrices, are sent to the fusion center. Secondly, after obtaining the result of track correlation, the state estimate and the covariance matrix about the same target of different sensors are fused according to the likelihood function of each model in the fusion center, and the model probabilities of the same target from different sensors are fused by using the new improved D-S combination rule of evidence which fuses the 3-dimensional(3-D) evidence together. Then the probability is used to update the target state estimate and the result is returned back to the local node. Therefore, more accurate state prediction of target is obtained. Finally, through simulations, the new algorithm is compared and analyzed with DIMM-MSJPDA-DS algorithm. Theoretical analysis and simulation results show that the new algorithm does well in tracking the strong maneuvering target. What’s more, through the whole algorithm, only a small amount of computation is involved. So it can be concluded that the new algorithm is an effective distributed interactive multi-model multi-sensor multi-maneuvering target tracking algorithm.

Key words: maneuvering target, interactive multiple model, generalized probabilistic data association algorithm, D-S combination rule of evidence

摘要: 为有效解决密集杂波环境下分布式多传感器多机动目标跟踪问题,提出了一种基于改进D-S证据组合规则的分布交互式多模型多传感器广义概率数据关联(DIMM-MSGPDA-IDS)算法。该算法首先对各局部节点均应用单传感器的IMM-GPDA算法跟踪多机动目标,并将其各模型的状态估计、协方差估计、模型概率、组合新息及其协方差矩阵等滤波结果送至融合中心;在航迹关联判决结束后,融合中心根据各模型对应似然函数的大小融合不同传感器关于同一目标的模型状态估计及其协方差矩阵,并提出利用三维(3-D)证据进行直接融合的改进D-S算法对来源于同一目标的不同传感器的各模型概率进行有效融合,然后依此概率来更新各目标的状态估计并反馈至各局部节点,使之获得更为精确的状态预测;最后,将该算法与基于D-S证据组合规则的分布交互式多模型多传感器联合概率数据关联(DIMM-MSJPDA-DS)算法进行仿真对比分析。理论分析和仿真结果表明,该算法能够很好地对强机动目标进行跟踪,且其计算量相对较小,是一种有效的分布交互式多模型多传感器多机动目标跟踪算法。

关键词: 机动目标, 交互式多模型, 广义概率数据关联算法, D-S证据组合规则