基于神经网络补偿的多传感器航迹融合

针对多传感器环境的条件提出了一种基于神经网络补偿的航迹融合方法.各传感器的测量值用线性卡尔曼滤波器进行处理并将获得的局部航迹传送到融合中心.首先对局部航迹进行融合,然后引入神经网络来减少因共同过程噪声而导致的融合估计误差,其中神经网络采用Dan Si-mon提出的网络结构,并对神经网络权值的优化采用无痕卡尔曼滤波(UKF).仿真结果表明,这种融合方法对跟踪具有过程噪声的目标非常有效,而且过程噪声发生变化时该方法仍是有效的,从而使得它在很多实际应用中具有潜在的价值.

Track Fusion with NN Compensated in a Multi-sensor Environment

The aim of this paper is to propose a track fusion method with NN compensated in a multi-sensor environment.The measurements of sensors tracking the same target are processed by local linear Kalman filters.The outputs of the local trackers are sent to the central node.In this node,simple Fusion is performed to the local tracks,then neural network is introduced to reduce the fusion estimation error due to the effect of common process noise.The neural network architecture is introduced just the same as Dan Simon brought forward,but the optimization of the network weights is using the unscented Kalman filter,which is computationally more efficient.The simulation results show that the fusion algorithm tracks the target with process noise very well,and it still performs well as the modelis varying,and it has a potential value in many real applications.