针对GPS/DR定位过程中采用卡尔曼滤波时噪声的统计特性与实际不符,提出采用小波神经网络嵌入到卡尔曼滤波,来实现自适应调整噪声协方差矩阵.通过对基于小波神经网络的自适应卡尔曼滤波辅助的GPS/DR定位系统进行仿真,结果表明既能有效抑制发散,又能有效提高定位精度.
GPS /DR positioning using Kalman filter has a large margin of error after a period of time,even divergent. This paper uses wavelet neural network in the Kalman filter to adjust the noise covariance matrix. Simulation results of mobile robot GPS/DR positioning system show that the divergence can effectively suppress,and can effectively improve the positioning accuracy.