论文部分内容阅读
针对车辆GPS/DR组合导航系统在GPS信号被遮挡时无法完成DR“零点更新”的问题,提出了基于BP神经网络的DR位置误差预测模型来解决该问题。在GPS有效时,该算法采用基于平稳小波变换的扩展卡尔曼滤波器对GPS/DR信号进行数据融合得到车辆实时的精确位置,与经平稳小波变换软阈值模平方去噪法处理的DR位置数据进行平稳小波多尺度比较获得DR位置误差;然后用BP神经网络建立DR位置误差预测模型,为了提高所用网络的泛化能力,采用了贝叶斯正则化规则训练网络。在GPS失效时,利用已建立的预测模型预测DR位置误差来修复DR位置数据,实现车辆行驶在复杂路径下的实时精确导航定位。仿真表明,该算法对车辆GPS/DR组合导航系统有效。
Aiming at the problem that vehicle GPS / DR integrated navigation system can not finish DR “zero update ” when GPS signal is occluded, a DR position error prediction model based on BP neural network is proposed to solve the problem. When GPS is valid, this algorithm uses the extended Kalman filter based on the stationary wavelet transform to get the real-time and accurate position of the GPS / DR signal and the DR position data Then the DR neural network is used to establish the prediction model of DR position error. In order to improve the generalization ability of the used network, the Bayesian regularization rule training network is adopted. In the event of GPS failure, the established prediction model is used to predict the DR position error to repair the DR position data, so as to realize real-time accurate navigation and positioning of the vehicle traveling in complicated path. Simulation results show that this algorithm is effective for GPS / DR integrated navigation system.