基于DKF的IMU误差预测算法

作为导航领域常用的组合导航方式,全球导航卫星系统(GNSS)/惯性导航系统(INS)组合导航在GNSS信号失锁后,由于惯性测量单元(IMU)误差随时间迅速积累,其定位结果会偏离载体真实位置,导航精度下降.针对此问题,提出了一种长短期记忆网络(LSTM)辅助的算法,称之为深度卡尔曼滤波(DKF)算法.DKF算法的核心思想是使用LSTM训练IMU误差模型,然后通过训练出的模型预测IMU误差,最后将预测的IMU误差代入IMU数据以校正导航结果.仿真结果表明:在200s测试数据上,DKF算法将误差从1.1537m/s降低到0.3746m/s.与平均预测、卡尔曼预测和最小二乘估计等方法相比,DKF算法的误差最小,具有更优越的导航性能.

IMU Error Prediction Algorithm Based on DKF

The combined navigation method of Global Navigation Satellite System(GNSS)/Inertial Navigation System(INS) is commonly used in the navigation field. After the GNSS signal loses, due to the rapid accumulation of Inertial Measurement Unit(IMU) errors over time, the positio-ning results will deviate from the true position of the carrier and the navigation accuracy will decrease. To solve this problem, a Deep Kalman Filter(DKF) algorithm based on Long Short-Term Memory(LSTM) is proposed. The core idea of the DKF algorithm is to use LSTM to train the IMU error model, and then predict the IMU error through the trained model, and finally bring the predicted IMU error into the IMU data to correct the navigation results. The simulation results show that the DKF algorithm reduces the error from 1.1537m/s to 0.3746m/s for the 200s test data. Compared with the methods of average prediction, Kalman prediction and least square estimation, the DKF algorithm has the smallest error and advanced navigation performance.