BP神经网络在车辆组合导航中的应用研究.pdf

331 Vol. 33 No. 1 20112 PI ZO L CTRICS ACOUSTOOPT ICS Feb. 2011 :2011) BP 李 红连 ( , 610106) :GPS/ DR GPS DR , BP DR GPS , GPS/ DR , DR DR ; BP DR , , GPS , DR DR , , GPS/ DR :GPS/ DR ; BP ;;; :V249 :A Research on Application of BP Neural Network to Vehicle Integrated Navigation System LI Honglian ( School of lectronic Information ngineering,Chengdu U niversity, Chengdu 610106, China) A stract: It is difficult to do dead reckon (DR) position! zero update! for vehicle GPS/ DR integrated navigation system in case of GPS signal blockage, a new DR position algorithm is put forward based on the backpropagation ( BP) neural netw ork. Firstly,DR position data is denoised w ith stationary w avelet transformation ( SWT) model square root softthreshold denoising algorithm,and the vehicle precise position data is acquired w ith extended Kal man filter( KF) data fusion algorithm based on SWT during the presence of GPS signal. The algorithm compares the tw o position data at different SWT decomposition level, reconstructs the SWT coefficients and the DR position error data is acquired. A BP network model to mimic DR position error property is trained w ith backpropagation al gorithm. To improve generalization of the BP netw ork, the backpropagation algorithm is improved by Bayesian regu larization ( BR) theory in this paper.During GPS outages, DR position error data is predicted w ith the model, and the algorithm provides precise position for the vehicle GPS/ DR integrated navigation system through DR position er ror data updating DR position data in complex running trajectory. The simulation results