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

Hydrodynamic process parameter calculation method based on double-flow neural network and application

A fluid dynamics and neural network technology, applied in the field of fluid dynamics process parameter calculation, to achieve the effect of sufficient mining of spatial structure feature information, rich features, and integrity assurance

Pending Publication Date: 2022-05-03
OCEAN UNIV OF CHINA
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] Aiming at the deficiencies in the prior art, the present invention provides a multi-task-based dual-stream neural network partial differential equation discretization method, which models and predicts the finite difference coefficients for the upstream and downstream respectively, uses the multi-view space fusion module to fuse the predicted coefficients, and details The granularity of the description of the physical process makes the mining of spatial structure feature information more sufficient, and the information contained is more abundant, which reduces the prediction error, improves the accuracy of the prediction coefficient, and solves the problem of low accuracy of the prediction coefficient in the existing technology

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
  • Hydrodynamic process parameter calculation method based on double-flow neural network and application
  • Hydrodynamic process parameter calculation method based on double-flow neural network and application
  • Hydrodynamic process parameter calculation method based on double-flow neural network and application

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0058] combine Figure 1-Figure 3 , which introduces the calculation method of fluid dynamics process based on two-stream neural network, including the following steps:

[0059] Step 1. Preprocess the data input into the fluid dynamics process parameter calculation model: create the Burgers equation, select an appropriate discrete format for discretization and calculate the true velocity value. specific:

[0060] First create the Burgers equation, and then use the fifth-order WENO format to calculate the Burgers equation to obtain the high-precision true velocity value of the Burgers equation as a baseline. For the WENO format, the following details:

[0061] The schematic diagram of the fifth-order WENO format is as follows image 3 As shown, the fifth-order WENO scheme is based on a six-point template S, which is divided into three subtemplates {S1, S2, S3}, with four nodes on each subtemplate. The steps are as follows: ①Determine the grid pedestal points: 6 pedestal poin...

Embodiment 2

[0116] This embodiment 2 provides an application of a fluid dynamics process calculation method based on a double-stream neural network, which is realized based on the fluid dynamics process calculation model of the present invention and the method described in embodiment 1, by inputting dynamic parameters into the fluid Calculation model of the dynamic process, the fluid dynamic field can be obtained by simulation, and the transient change of a certain space point can also be obtained.

Embodiment 3

[0118] This embodiment provides a specific application of a fluid dynamics process calculation method based on a double-stream neural network, which is used in aerospace. The velocity value and the finite difference coefficient in the Burgers equation are calculated by the method to determine the shape of the fluid, which is used for the wing Shape development and design. At present, the application of fluid mechanics in the field of aerospace has become more and more important, the most obvious of which is the manufacture of wing shape. There are various shapes of aircraft wings, among which the flat wing section is equivalent to the section of a kite, which generates lift by the angle of attack; the bird wing section is mostly used in early aircraft; the slightly flat wing section with an upper arch and a lower aerodynamic force is particularly good Lift is mostly used in subsonic aircraft and so on. The shape of the wing needs to be considered according to the shape of the...

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 fluid dynamic process parameter calculation method based on a double-flow neural network and application, and the method comprises the steps: inputting a real speed value of a Burgers equation into a double-flow neural network of a fluid dynamic process parameter calculation model, and respectively predicting finite difference coefficients of an upwind direction and a downwind direction; and then fusing the finite difference coefficients of the upwind direction and the downwind direction by using a multi-view space fusion module, generating a finite difference coefficient at a central point by combining the weight value at each point, then calculating a space derivative and a time derivative, and further predicting a speed value through a time derivative value. According to the method, the accuracy of the difference coefficient is improved.

Description

technical field [0001] The invention belongs to the technical field of neural networks, in particular to a calculation method and application of fluid dynamics process parameters based on a double-stream neural network. Background technique [0002] The discretization method of dynamic equations in physical processes is an important method for simulating fluid dynamics and high-energy physical phenomena. Discretization work is to use finite difference, finite volume and other discretization methods to replace the continuous calculation area with a series of grid nodes to solve the problem that partial differential equations cannot obtain analytical solutions. However, the problems faced by these traditional discretization methods are: 1. Discretization can only be performed on fine-grained grids, and the accuracy of approximate solutions on coarse grids is not high. Only in the lattice can the integral conservation be satisfied. 2. The speed of numerical solution is slow a...

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 Applications(China)
IPC IPC(8): G06F30/27G06F30/28G06F30/23G06N3/04G06N3/08G06F113/08G06F119/14
CPCG06F30/28G06F30/27G06F30/23G06N3/04G06N3/08G06F2113/08G06F2119/14
Inventor 聂婕耿浩冉王成龙王京禹陈昊杨启成
Owner OCEAN UNIV OF CHINA
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