带BP神经网络的卡尔曼滤波算法在GPS／SINS组合导航系统中的应用

Application of Neural Network BP Embedded In Kalman Filtering algorithm Based on Integrated Navigation System of Navigation of GPS/SINS

针对带权值调整的Kalman滤波器,运用L-M的BP算法,将BP神经网络嵌入该滤波器中,与BP神经网络滤波器相比,减小了层数,提高了网络训练速度及精度.以GPS/SINS组合导航系统为例进行了仿真,结果既能抑制滤波发散,又能提高滤波精度.

In relation to the Kalman filter adjusting the weighting, the method of L-M BP algorithm is used and Neural Network BP is embedded in Kalman filter. Comparing with Neural Network BP filter, not only the layers are decreased, but also the training velocity and accuracy of BP network are improved. The result of computer simulation in the GPS/SINS integrated navigation system shows the algorithm is useful and effective with high effectiveness.