Machine Learning Model for Aerodynamic Flow Velocity Field Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional aerodynamic analysis methods fail to accurately model flow velocity fields when fluid inflow velocity is variable, leading to decreased learning accuracy due to the lack of inflow velocity information in shape models and the uniform learning of similar flow velocity fields as distinct data.
Innovation Solution
The method incorporates the Reynolds number into the object shape model to estimate flow velocity fields, considering the inflow velocity and shape characteristics, and uses machine learning to create training data associating boundary layer, diffusion, and wake flow angles with flow velocity fields, enabling the estimation of flow velocity fields under varying simulation conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional aerodynamic analysis uses shape models without inflow velocity information, then the model construction is simple, but the learning accuracy decreases when inflow velocity is variable
Solution Approach 1:
The patent transforms the object shape model by incorporating the Reynolds number as an additional parameter. This allows the model to capture the relationship between shape characteristics and flow velocity fields under varying inflow conditions, resolving the contradiction between model simplicity and estimation accuracy.
2Productivity
If uniform learning is applied to similar flow velocity fields, then the learning process is efficient, but overfitting occurs and prediction accuracy decreases
Solution Approach 1:
The patent introduces local quality differentiation by incorporating boundary layer position, diffusion range, and wake flow characteristics as distinct features in the training data. This allows the machine learning model to learn specific local flow characteristics while maintaining overall learning efficiency, preventing overfitting.
3Measurement precision
If aerodynamic simulation is executed for each shape variation, then the flow velocity field estimation is accurate, but the computational cost and time increase significantly
Solution Approach 1:
The patent creates a machine learning model that copies the essential flow velocity field characteristics from simulation data without requiring actual simulation execution for each shape variation. The model learns from training data including boundary layer position, diffusion range, and wake flow parameters, enabling rapid prediction without repeated simulations.
4Measurement precision
If detailed flow velocity field data is collected for training, then the prediction accuracy improves, but the data preparation complexity and computational resources increase
Solution Approach 1:
The patent extracts key flow velocity field characteristics (boundary layer position, diffusion range, wake flow characteristics) from detailed simulation data to create a simplified yet representative training dataset. This extraction approach maintains prediction accuracy while reducing data preparation complexity and computational resource requirements.
Data Source
AI summary
A machine learning method implemented by a computer includes: acquiring simulation conditions including a shape of an object and an inflow velocity of fluid; identifying, based on the shape of the object and the inflow velocity that have been acquired, a position of a boundary layer with respect to the object, a diffusion range of the fluid, and a flow velocity diffusion range of a wake flow of the fluid; creating training data associating the position of the boundary layer, the diffusion range of the fluid, and the flow velocity diffusion range of the wake flow of the fluid that have been identified with a flow velocity field under the simulation conditions; and generating a model that estimates the flow velocity field from the simulation conditions by using the training data.