Back to results
US 11,531,795 B2Granted
Apparatus for optimizing flow analysis and method therefor
AssigneeDosan Enerbility Co., Ltd.
InventorsJeehun Park, Hyunsik Kim, Jaehyeon Park, Sangjin Lee
Life of this patent
Estimating the term…
Abstract
A flow analysis apparatus is provided. The flow analysis apparatus includes a flow analyzer configured to derive a plurality of output signals by performing flow analysis for a plurality of cells by using a flow analytic model for simulating numerical analysis by Computational Fluid Dynamics (CFD) with respect to a plurality of cells that divide a space around a component, and an analysis optimizer configured to perform optimization for the plurality of output signals.
Claims10
A flow analysis apparatus comprising:
- a processor and a memory storing computer program commands, the computer program commands when executed by the processor implement the steps of:
- (a) acquiring training data for an artificial neural network comprising;
- storing, in the memory, a plurality of input signals with respect to a plurality of cells that divide a space around a structural component wherein the plurality of input signals includes a laminar flow viscosity or a turbulent conduction and;
- producing, by the processor, a plurality of output signals at an initial stage of fluid flow corresponding to each of the plurality of input signals by performing Computational Fluid Dynamics (CFD) numerical analysis a predetermined number of times, wherein the plurality of output signals includes a density, a momentum, or an internal energy;
- (b) training, by the processor, parameters of a first artificial neural network model by inputting the plurality of input signals and the plurality of output signals to the first artificial neural network model, wherein the first artificial neural network model predicts an input signal at a steady state of the fluid flow;
- (c) generating, by the processor, a plurality of predicted input signals using the first artificial neural network model;
- (d) training, by the processor, parameters of a second artificial neural network model by inputting the plurality of output signals and the plurality of predicted input signals to the second artificial neural network model, wherein the second artificial neural network model predicts an output signal at the steady state of the fluid flow;
- (e) generating, by the processor, a plurality of predicted output signals using the second artificial neural network model;
- and
- (f) performing optimization for the plurality of predicted output signals.
7 more elements
A flow analysis method for predicting dynamics of fluids around a structural component using artificial neural network models, thereby reducing time spent in design and analysis of the structural component, comprising:
- (a) acquiring training data for an artificial neural network comprising;
- storing, in the memory, a plurality of input signals with respect to a plurality of cells that divide a space around the structural component wherein the plurality of input signals includes a laminar flow viscosity or a turbulent conduction and;
- producing, by the processor, a plurality of output signals at an initial stage of fluid flow corresponding to each of the plurality of input signal by performing Computational Fluid Dynamics (CFD) numerical analysis a predetermined number of times, wherein the plurality of output signals includes a density, a momentum, or an internal energy;
- (b) training, by the processor, parameters of a first artificial neural network model by inputting the plurality of input signals and the plurality of output signals to the first artificial neural network model, wherein the first artificial neural network model predicts an input signal at a steady state of the fluid flow;
- (c) generating, by the processor, a plurality of predicted input signals using the first artificial neural network model;
- (d) training, by the processor, parameters of a second artificial neural network model by inputting the plurality of output signals and the plurality of predicted input signals to the second artificial neural network model, wherein the second artificial neural network model predicts an output signal at the steady state of the fluid flow;
- (e) generating, by the processor, a plurality of predicted output signals using the second artificial neural network model;
- and
- (f) performing optimization for the plurality of predicted output signals, wherein the performing the optimization comprises:
- removing, by a filter, noise in each of the plurality of predicted output signals;
- optimizing, by a primary optimizer, the predicted output signal from which the noise has been removed primarily;
- and
- optimizing, by a secondary optimizer, the primarily optimized output signal secondarily.
10 more elements
Dependencies are inferred from the claim text.