Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A method and device for generating fluid density field based on deep learning

A technology of fluid density and deep learning, applied in neural learning methods, biological neural network models, instruments, etc., can solve problems such as low calculation accuracy, low resolution, and noise sensitivity

Active Publication Date: 2021-07-30
ZHEJIANG UNIV
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

>[0022] In order to solve the problems existing in the background technology, the object of the present invention is to provide a method and device for calculating the density field from the background schlieren image, solving the low calculation accuracy and resolution existing in the current calculation density field Low and noise sensitive issues

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A method and device for generating fluid density field based on deep learning
  • A method and device for generating fluid density field based on deep learning
  • A method and device for generating fluid density field based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0104] 1. Generate a data set

[0105] The principle of data generation is as follows Figure 4 shown

[0106] (1) Construct a uniform density field 302 in advance, and perform different forms of heat source heating to change the density of the flow field under certain working conditions, and then use CFD simulation calculations to calculate a series of non-uniform density fields 306;

[0107] (2) Use zemax and other optical software to simulate the light paths of the light rays of the background pattern 301 passing through the uniform density field 302 and the above-mentioned non-uniform density field 306 with density gradients, as shown in figure 2 As shown, image sets are composed of background texture images at different camera positions at a preset camera position (the number of acquisition positions N≥2), and a uniform density background texture image set 303 and a non-uniform density background texture image set 305 are respectively obtained;

[0108] (3) Using the a...

Embodiment 2

[0116] The difference between the second embodiment and the first embodiment is that the input of the neural network and the generation of the data set are different.

[0117] The data set adopts the method of generating the second data set, and in step (3), an additional background pattern 301 is added as input to form the second data set.

[0118] The network structure of the second embodiment is as Figure 6 As shown, there is an input layer and an output layer. The input layer input image sets 303 and 305 and the image after the background pattern 301 are concatenated, its dimension is (256, 256, 7, 1), and then a set of 3D convolution is set The first layer extracts the features of the spliced ​​images; after that, a set of deconvolution layers is set up to amplify the features, and finally the high-resolution and high-precision density field is restored through the output layer.

[0119] In practical applications, the two background texture image sets 303 and 305 and th...

Embodiment 3

[0122] 1. Generate a data set

[0123] The principle of data generation is as follows Figure 4 shown.

[0124] (1) Construct a uniform density field 302 in advance, and perform different forms of heat source heating to change the density of the flow field under certain working conditions, and then use CFD simulation calculations to calculate a series of non-uniform density fields 306;

[0125] (2) Use zemax and other optical software to simulate the light paths of the light rays of the background pattern 301 passing through the uniform density field 302 and the above-mentioned non-uniform density field 306 with density gradients, as shown in figure 2 As shown, image sets are composed of background texture images at different camera positions at a preset camera position (the number of acquisition positions N≥2), and a uniform density background texture image set 303 and a non-uniform density background texture image set 305 are respectively obtained;

[0126] (3) For the un...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a method and device for generating a fluid density field based on deep learning. Construct a uniform density field and a non-uniform density field; respectively simulate the refraction light path of the light passing through the background pattern in the uniform density field and the non-uniform density field, collect images at different camera positions to form an image set, take the image set as input, and use the non-uniform density field as input The density field is used as a label to form a training sample; the above steps are repeated in different situations to construct a training data set; a neural network for calculating the density field is established and trained; the background image to be tested is input into the neural network model to obtain the final result. The present invention utilizes the double information of the background pattern and the three-dimensional density field to obtain results, has high precision and high resolution, can effectively eliminate the influence of noise, has strong algorithm robustness, and solves the problem of insufficient training data in the neural network training process question.

Description

technical field [0001] The invention relates to a fluid data processing method and device in the field of fluid experiment flow measurement and display, in particular to a method and device for generating a fluid density field based on deep learning. Background technique [0002] At present, the techniques used for density field measurement mainly include: Schlieren method, shadow method, and interferometry. [0003] The shadow method records the deflection position difference, which reflects the change of the refractive index gradient, that is, the second derivative of the refractive index. The shadow method is often used in places where the density gradient changes greatly, suitable for large scales, and relatively low material requirements. Lowest cost. [0004] The Schlieren method records the deflection angle difference, which reflects the gradient of the refractive index, that is, the first derivative of the refractive index. The Schlieren system is relatively simple...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06F30/28G06F30/27G06N3/04G06N3/08G06F113/08
Inventor 高琪
Owner ZHEJIANG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com