卫星与雷达位置数据自适应关联

卫星与雷达间的数据关联能够实现海上预警探测过程中由大范围预警向精细跟踪过渡转换,但传统关联模型关联速度慢且难以适应舰船编队目标的非刚性变换、虚警漏报等情形。对此，提出一种卫星与雷达位置数据自适应关联模型。首先，采用多层神经网络提取卫星数据和雷达数据的整体差异参数。然后，将参数通过位移变换估计网络实现两类信源目标的匹配,解决不同信源间的时间间隔和定位误差导致的空间位置差异。最后，对匹配后的目标进行关联判决。仿真实验结果表明,该模型关联速度快,精度高,能够实时处理大规模多源数据关联任务。同时在应对非刚性变换、定位误差、虚警漏报等场景下有较好的鲁棒性。

Adaptive association for satellite and radar position data

The data association between satellite and radar can realize the transition from large-scale early warning to fine tracking in the process of the early warning detection. However, the traditional association model has a slow association speed and is difficult to deal with the non-rigid transformation, false alarm and missing detection of ship formation targets. In this regard, an adaptive association model is presented for satellite and radar position data. Firstly, the multi-layer network (MLN) is used to extract the global difference parameters of the satellite data and the radar data. Then, in order to solve the spatial position error caused by time interval and positioning error, the parameters are put into the network of displacement transformation estimation to match the targets from the two types of sources. Finally, an association judgment is accomplished for the matched targets. Simulation and experimental results show that the proposed model achieves good performance both on speed and accuracy, and can handle the large-scale multi-source data association in real-time application. Besides, the model has good robustness under the conditions of non-rigid transformation, positioning error, false alarm and missing detection.