Geometry parameters recognition method for underwater target model based on Kriging surrogate model

Abstract

ObjectivesUnderwater target parameter recognition can provide the basis for target classification and recognition. To this end, an underwater target parameter recognition method based on the Kriging surrogate model is proposed. MethodsFirst, the low-frequency acoustic radiation of the structure surface and the acoustic sensitivity of an underwater target model covered with an acoustic layer are analyzed under propeller and main and auxiliary engine excitation. A low-frequency acoustic radiation power surrogate model is then established using the above analysis results, and a sample space consisting of acoustic radiation response features and target parameters is constructed on the basis of the surrogate model. Finally, a target parameter recognition surrogate model is established on the basis of the sample space, and test points are selected for model validation. ResultsThe results show that the actual target parameter values of the test samples match well with the predicted target parameter values of the constructed surrogate model. The finite element method (FEM) and boundary element method (BEM) can realize the low-frequency acoustic radiation analysis of viscoelastic damping structures by considering the frequency-dependence characteristics of the damping material, and are able to solve the problem that commercial software may not process vibration result files in large quantities. The main target parameters affecting the low-frequency vibro-acoustic radiation of the underwater target model are the model length, maximum radius of the target, model thickness and acoustic layer thickness. ConclusionsThe underwater target parameter recognition method based on a surrogate model proposed herein is able to predict the main target parameter values of the underwater target model by analyzing its acoustic radiation line spectral features.

Keywords

• surrogate model
• acoustic radiation
• underwater target parameters recognition
• acoustic sensitivity
• viscoelastic damping