Research On Integrated Navigation Algorithm Based On CKF In Complex Environment

With the rapid development of science and technology,users have higher requirements for real-time high-precision positioning information.The integrated navigation system can overcome the disadvantages of single system and obtain reliable positioning information,so the integrated navigation system has been widely studied and applied.However,vehicles and other carriers operate in complex environments such as cities with tall buildings,and satellites are often blocked by tall buildings,tunnels,overpasses and vegetation,so the use of traditional Kalman filtering algorithms will cause the integrated navigation system's positioning results to be less accurate or even scattered.In order to provide high precision navigation information for GNSS/INS integrated navigation system when GNSS signals are blocked for a long time,an information fusion algorithm of interactive multiple model adaptive robust cubature Kalman filter assisted by Gated Recurrent Unit neural network was proposed.Firstly,adaptive CKF and robust CKF,are proposed to solve the problem of accuracy degradation caused by the uncertainty of the system model and the uncertainty of the statistical properties of the noise when using the combined GNSS/INS navigation system for positioning.Then,the adaptive factor is introduced to adjust the covariance matrix of state prediction to reduce the impact of dynamic model errors.In order to overcome the statistical error of measurement noise,an online adjustment method of CKF algorithm based on robust estimation is proposed.Secondly,the two improved filters mentioned above are fused using the interacting multiple model,and the final system state estimation is the probabilistic weighted sum of the results of the two sub-filters.By combining the advantages of the two improved filters,the method can effectively suppress the interference of the uncertainty of the system model and measurement noise statistics to the filter solution.Finally,the GRU neural network is introduced into the GNSS/INS system,which includes two modes of training and prediction.When GNSS signal can be received,IMM-ARCKF is used for filtering and the GRU neural network works in the training mode.When GNSS signal is lost,the GRU neural network predicts GNSS position increment.After the position increment integration,the proposed IMM-ARCKF is used for filtering and solving.In order to evaluate the effectiveness of IMM-ARCKF algorithm,simulation comparison experiments as well as real vehicle sports experiments are designed.In chapter 3,the experimental results show that the IMM-ARCKF can effectively suppress the interference of the system model uncertainty and the statistical uncertainty of measurement noise on the filtering solution.In chapter 4,the experimental results show that GRU/IMM-ARCKF algorithm can effectively improve the positioning accuracy and reliability of GNSS/INS system during GNSS interruption.