Underwater mechanism cross-scale flow field feature prediction method based on improved Unet network

Abstract

The invention provides an underwater mechanism cross-scale flow field feature prediction method based on an improved Unet network. The method comprises the following steps: constructing a batch of training set sample data sets; constructing an improved Unet network for cross-scale flow field feature prediction of the underwater mechanism based on the batch of training set sample data sets; and constructing a loss function model in combination with a real label corresponding to each batch of training set sample data sets, and training the Unet network to obtain a cross-scale flow field feature prediction model of the underwater mechanism. The improved Unet prediction model is provided, cross-scale flow field change prediction is achieved through depth feature extraction and feature enhancement fusion, prediction precision is improved, and the problem that a traditional time sequence method is low in precision is solved; and the model prediction capability can be enhanced through online learning of the flow field data set of each working condition, and second-level high-quality time sequence prediction output of the flow field at an unknown moment is realized based on a small amount of sample training.

Description

technical field [0001] The invention relates to the field of combination of fluid mechanics simulation and artificial intelligence, in particular to a method for predicting cross-scale flow field characteristics of an underwater mechanism based on an improved Unet network. Background technique [0002] With the continuous development of marine resources and organisms, the role of various underwater institutions is becoming more and more important, especially underwater institutions such as submarines and AUVs (Underwater Autonomous Vehicles) have important scientific and military significance. The development of Computational Fluid Dynamics (CFD) has made it convenient for people to study the performance of various underwater mechanisms. However, for complex underwater structures, there are still problems of time-consuming modeling and high computational costs. [0003] The continuous development of artificial intelligence methods provides new solutions to engineering probl...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art, provide a method for predicting the cross-scale flow field characteristics of underwater mechanisms based on the improved Unet network, and obtain the cross-scale prediction of the motion position of the underwater mechanism and the time series results of the flow field in real time , and solved the problem of unclear flow field image predicted by the current technology. The actual measurement results show that the method in this paper realizes the cross-scale accurate time series prediction of the flow field characteristics of the underwater mechanism, and the mse error is reduced by two orders of magnitude compared with the existing technology.

Images