An Iterative Doppler Velocity Log Error Calibration Algorithm Based on Newton Optimization

Abstract

The performance of the inertial navigation system/Doppler velocity log (INS/DVL) is mainly degraded by DVL error. An optimization-based algorithm is widely used and effective on calibrating DVL error. In this algorithm, the loss function established is a quadratic function of DVL attitude error; however, the existing optimization-based calibration method is just of first-order generally, so that it is not able to reach the best calibration efficiency. In order to improve the efficiency, an iterative calibration algorithm based on the second-order Newton optimization is proposed innovatively to calibrate the DVL error. In addition, the Wolfe principle is introduced to limit its loss function's oscillation around the extreme point to improve the efficiency and accuracy of the Newton optimization. The excellent efficiency and accuracy of the proposed calibration scheme are validated by simulation and field tests.