Navigation Algorithm Based on PSO-BP UKF of Autonomous Underwater Vehicle

The development of precise localization and autonomous navigation for Autonomous Underwater Vehicles (AUVs) is fundamental to reach the high level of performance required by complex underwater tasks. One of the main factors affecting the accuracy of AUV navigation systems is the algorithm used to estimate the vehicle motion. In order to achieve high accuracy of navigation, Unscented Kalman Filter (UKF) has been proposed by some scholars to overcome the limitations of EKF, such as accumulation of error and the calculation of the Jacobian matrixes. A novel windowing based Adaptive Unscented Kalman Filter (AUKF) is presented to address the performance degradation and divergence of the standard UKF due to unknown or inaccurate statistical characteristics of system noise. Nevertheless, sometimes AUKF can't significantly improve navigation accuracy. This paper presents an improved UKF algorithm by combining the Particle Swarm Optimization (PSO) PSO and Back Propagation neural network (BP) algorithm (PSO-BP UKF), which can use the prediction value of PSO-BP to correct the error of UKF. Besides, the algorithm based on PSO-BP UKF is evaluated by a range of real data in the sea trial, and is significantly superior to that of the standard UKF and AUKF, leading to improved calculation precision of the integrated navigation system.