A Sound Velocity Profile Stratification Method Based on Maximum Density and Maximum Distance Clustering

Abstract

In the field of deep-sea positioning, this paper aims to enhance accuracy and computational efficiency in positioning calculations. We propose an improved method based on layered clustering of sound velocity profiles, where the profiles are stratified according to maximum distance and maximum density. Subsequently, a secondary curve fitting is applied to the stratified data. Ultimately, the underwater positioning is conducted using the sound velocity profiles’ post-layered fitting. We compare our approach with traditional methods such as k-means clustering, layered clustering, and gradient-based stratification. Experimental results demonstrate that, in the application scenario of a USBL system with a transducer tilted at 30°, and under the premise of autonomously controlling the number of layers, our method significantly improves positioning accuracy.