Research On GNSS/INS Integrated Navigation Model Aided By BP Neural Network

With the rapid development of science and technology,high-precision positioning and navigation technology has become one of the current research hotspots.A single navigation system is difficult to meet the needs of users,and integrated navigation is an inevitable trend in the development of navigation technology.GNSS/INS is currently the most widely used integrated navigation.It uses the stability and high precision of the satellite navigation system to compensate for the increase in errors of the inertial navigation system over time,and uses the short-term high precision of the inertial navigation system to compensate for the satellite When the navigation receiver is interfered,the signal is lost,the error increases,etc.,so as to achieve complete high-precision navigation and positioning.However,in a complex environment,the lack of GNSS signals can easily accumulate INS errors and reduce positioning accuracy.In order to solve this problem,this article focuses on the GNSS/INS integrated navigation in the loose combination mode and deeply studies a variety of fusion algorithm models,and proposes GNSS/INS model assisted by BP neural network.The main research contents and results are as follows:(1)This article first introduces the principle of GNSS positioning and its error correction,then explains the principle of INS,and derives the INS coordinate conversion and update algorithm.The standard Kalman filter algorithm under loose combination mode is studied,and it is pointed out that the Kalman filter algorithm is difficult to apply to nonlinear systems.Aiming at the limitations of Kalman filtering,extended Kalman filtering and unscented Kalman filtering are introduced.Finally,the superiority of BP neural network assisted GNSS/INS under the condition of GNSS signal loss is explained.(2)In this study,a complete set of vehicle-mounted experimental platform was built,and the design experiment verified the positioning accuracy and applicability of the standard Kalman filter,extended Kalman filter and unscented Kalman filter algorithm in the loose combination mode.The experimental results show that the standard Kalman filter algorithm has great limitations in solving nonlinear problems,and the positioning result is poor;the extended Kalman filter algorithm is second;the unscented Kalman filter has higher positioning accuracy,and its three-axis position error mean square The roots are all within 0.6m,and the root mean square of the three-axis speed errors are all within 0.04m/s.The unscented Kalman filter algorithm has shown strong superiority in GNSS/INS integrated navigation data fusion filtering.(3)Designed the GNSS/INS integrated navigation system model assisted by the BP neural network for the problem of GNSS signal missing.When the GNSS signal is normal,the integrated navigation system is filtered and calculated,and the INS data is used as the input value to complete the BP neural network.Training work,simulating GNSS signal loss conditions,using BP neural network predicted output value to compensate the system error,and comparing and analyzing the results of integrated navigation without BP neural network assistance.The experimental results show that the position error of the BP neural network prediction result is within 6m,The speed error is within 0.5m/s,indicating that the BP neural network can suppress the error divergence of the integrated navigation system to a certain extent,and ensure the overall accuracy of the integrated navigation system.Figure 25 Table 6 Reference 87...