基于神经网络辅助观测的连续组合导航算法研究

Research on a smart integrated navigation algorithm based on neural network-aided observation

针对因全球定位系统(GPS)信号失效导致捷联式组合导航系统SINS/GPS组合导航系统发散的问题,设计了一种基于神经网络辅助观测的智能组合导航算法。该方法在GPS信号有效时训练神经网络,当GPS失效后利用神经网络自主重构组合导航系统,将神经网络的输出信息作为观测量构建新的Kalman滤波器,以实现对捷联惯性导航系统误差的连续反馈校正,从而实现了高精度的连续导航。该方法得到了仿真验证,从仿真结果可以看出,在GPS短时失效的情况下,该方法有效抑制了姿态角、速度和位置的发散现象,提高了组合导航系统的精度和可靠性。

The paper presents a smart integrated navigation algorithm based on neural network-aided observation, aiming at the divergence problem of integrated navigation systems caused by global positioning system （GPS） outages. The method trains the neural network when the GPS is available, and it independently rebuilds the integrated navigation system using the neural network when GPS outages occur. The output of the neural network is utilized as the measurement to build a new Kalman filter, which is used to amend the error of the strapdown inertial navigation system （SINS）, and then the continuous navigation with high precision is realized. The simulation was carried out. The result demonstrated that the divergence of attitude, velocity and position were effectively controlled under this algorithm. The precision and reliability of integrated navigation systems were improved.