Research on Strong Tracking UKF Algorithm of Integrated Navigation Based on BP Neural Network

Abstract

Aiming at vehicles in the dense environment of tunnels, viaducts, mountainous areas, and high-rise buildings, GPS signals often suffer from short-term lock-out. A strong tracking unscented Kalman filter (STUKF) integrated navigation algorithm derived from Back Propagation neural network was proposed. This paper combines the idea of strong tracking filtering with the idea of unscented Kalman filtering, with the assistance of BP-neural network, and applies it to GPS/SINS integrated navigation with complementary advantages. The availability and reliability of the algorithm are tested by experimental simulation. Compared with the influence of BP neural network training before and after training on the accuracy of integrated navigation, the test results that this algorithm not only overcomes the shortcomings of GPS signal unlocking in harsh environment and Kalman filter fluctuates greatly in nonlinear environment, but also immensely improves the positioning precision of integrated navigation system, and provides new ideas for intelligent navigation fields such as unmanned vehicles and drones.