多传感器数据融合与航迹预测的模型设计

文中建立了功能完整的多传感器数据融合模型,使用最邻近数据关联(NNDA)算法提取目标航迹,采用三次样条插值实现时间配准,并将传感器性能作为权重应用到航迹对的融合中,使用加权平均的方法融合航迹对.引用回声状态网络(ESN)技术实现航迹的预测.基于雷达数据在不同运动场景中测试模型性能,对于匀速直线运动和机动转弯的目标,模型具有较高的跟踪和预测精度.

Design of multi-sensor data fusion model with trajectory prediction

Multiple sensors can improve system reliability,extend the space-time coverage and enhance information utilization.Multi-sensor data fusion is an important part of multiple sensor system.In this paper,a multi-sensor data fusion model was designed,with nearest neighbor data association（NNDA） algorithm for target tracking and with cubic spline interpolation method for time registration.Performance of the sensor was measured and used as the weight to achieve track fusion and echo state networks（ESN） was used for trajectory prediction.Based on the radar data,simulations were presented to test the performance under different motion scenarios,drawing the conclusion that for uniform rectilinear motion and turning the target,the system has a higher tracking and prediction accuracy.