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

Fluid simulation data prediction model based on LSTM

A fluid simulation and data prediction technology, which is applied in electrical digital data processing, design optimization/simulation, biological neural network model, etc.

Active Publication Date: 2021-01-05
SHANGHAI UNIV
View PDF4 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, these parameters are difficult to measure experimentally.

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
  • Fluid simulation data prediction model based on LSTM
  • Fluid simulation data prediction model based on LSTM
  • Fluid simulation data prediction model based on LSTM

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0053] An LSTM-based scroll compressor fluid simulation data prediction model is characterized in that: for the scroll compressor fluid, a combination of finite element simulation and LSTM data prediction is used to increase the fluid simulation speed, thereby improving scroll compression. The machine-fluid simulation process is characterized by:

[0054] (1) Perform a scroll compressor fluid simulation:

[0055] Firstly, a three-dimensional model of the fluid domain of the scroll compressor is established, and the dynamic mesh method based on the FLUENT software is used to perform a transient calculation of the three-dimensional flow field inside the scroll compressor. The internal transient flow field value at time, thus obtaining the distribution of internal pressure, temperature and velocity of the flow field under different working conditions;

[0056] (2) Establish three observation points at the Z=0.018m plane of the internal flow field of the scroll compressor, and ob...

Embodiment 2

[0062] This embodiment is basically the same as the first embodiment, and the special features are:

[0063] In this example, see figure 1 , use L1 regularization to reduce the complexity of the neural network model, and its loss function C is:

[0064]

[0065] In the above formula, the modified term sums the weights in the neural network, λ represents the regularization parameter, n is the number of samples contained in the training set, W represents the weight of the neuron, x is the input value of each neuron, and y is the Output value, α is the estimated value, L is the power of the estimated value, w is the weight of each neuron, and the partial derivative of C with respect to W can be obtained:

[0066]

[0067] In the above formula, sgn represents the symbolic function, C 1 is the original cost function, W represents the weight of the neuron; when W is greater than 0, the result is 1, and when W is less than 0, the result is -1; the update method of the weight ...

Embodiment 3

[0095] This embodiment is basically the same as the above-mentioned embodiment, and the special features are:

[0096] In this embodiment, as figure 1 As shown, an LSTM-based fluid simulation data prediction model includes the following steps:

[0097] Feed the time series data of the observation points into the LSTM neural network, first perform normalization processing, and map all the data to the interval 0-1;

[0098] Then continuously fine-tune the loss function and the regularized parameter values ​​in the LSTM layer during model training;

[0099] Finally, the present invention adopts the absolute mean square error as an index to measure the error performance of the neural network.

[0100] In this embodiment, preprocessing is performed on the collected thermodynamic time series data of observation points, and the original data is normalized by using a discrete normalization processing method;

[0101]

[0102] where max is the maximum value of the sample data, an...

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 provides a scroll compressor fluid simulation data prediction model based on LSTM. The method comprises the steps: calculating a three-dimensional transient flow field value of an internal working cavity of the scroll compressor during operation by using a method of calculating fluid dynamics; collecting the change conditions of temperature, pressure and speed values at the observation points along with time; finally, substituting the acquired time sequence data into an LSTM network for training, and predicting the change trend of the thermodynamic value of the observation pointalong with time by utilizing the trained prediction model.

Description

technical field [0001] The invention designs a flow field data prediction model, specifically a fluid simulation data prediction model based on LSTM, which is used to replace the finite element software to perform trend prediction on the thermodynamic data of observation points in the flow field, and then from the perspective of simulation optimization Shorten the design and development cycle of the entire fluid device. Background technique [0002] Scroll compressors have the advantages of small size, wide working pressure range, simple structure, high working efficiency and low noise and vibration, and have been widely used in the industrial field. The working process of the scroll compressor is a three-dimensional unsteady compressible viscous flow process. In the process of designing and verifying the scroll compressor, it is very important to obtain the state parameters during its working process. However, these parameters are difficult to measure experimentally. At ...

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
IPC IPC(8): G06F30/27G06F30/28G06N3/04G06N3/08G06F113/08G06F119/08G06F119/14
CPCG06F30/27G06F30/28G06N3/049G06N3/08G06F2113/08G06F2119/08G06F2119/14G06N3/045
Inventor 余尧饶家凯王秀梅
Owner SHANGHAI 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