Machine Learning-Assisted Prediction for Complex Fluid Flow Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing simulation algorithms for complex fluid flow problems require significant computational resources and time due to their complexity and large-scale discretization, necessitating High Performance Computing clusters.
Innovation Solution
Integrate a machine learning model trained on simulation results into the simulation process, using it to provide an approximate estimation for subsequent simulations, thereby reducing computational requirements and improving efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional simulation algorithms are used for complex fluid flow problems, then simulation accuracy is maintained, but computational time and resource consumption increase significantly
Solution Approach 1:
The patent creates a machine learning model that copies the input-output relationship of the traditional simulation algorithm. Instead of executing the complex simulation algorithm directly, the trained ML model produces approximate results that replicate the simulation's behavior, significantly reducing computational time while maintaining acceptable accuracy for prediction tasks
Solution Approach 2:
The patent substitutes the mechanical computation process of traditional simulation algorithms with a machine learning inference process. The ML model, once trained, performs predictions through neural network computations that are much faster than solving complex fluid flow equations, replacing the heavy mechanical simulation with a lightweight computational approach
2Measurement precision
If traditional simulation algorithms are used for complex fluid flow problems, then accurate results are obtained, but computational resources and complexity increase
Solution Approach 1:
The machine learning model captures the input-output mapping of the complex simulation algorithm, creating a simplified computational copy that reproduces simulation results without requiring the full computational machinery of the original simulation engine
Solution Approach 2:
The patent extracts the essential prediction capability from the complex simulation algorithm by training an ML model on simulation data. This separates the learning phase (where complexity is concentrated) from the prediction phase (where simplicity is achieved), allowing fast predictions without the original simulation's computational burden
3Productivity
If machine learning model is integrated into simulation process, then computational efficiency is improved, but model training requirements increase
Solution Approach 1:
The patent performs the computationally intensive work in advance by training the machine learning model on simulation data before actual predictions are needed. This preliminary training phase consolidates the computational effort, allowing subsequent predictions to be made quickly without repeating the heavy computation
Solution Approach 2:
The patent divides the computational process into two distinct phases: an offline training phase where the ML model learns from simulation data, and an online prediction phase where the trained model provides fast results. This segmentation allows heavy computation to be performed once during training, while operational predictions remain lightweight and efficient
Data Source
AI summary
Embodiments of the present disclosure provide a method, an electronic device, and a computer program product for determining a prediction result. The method includes acquiring a first operating condition data set and outputting a first output result corresponding to the first operating condition data set through a machine learning model. The machine learning model is obtained by training on a second operating condition data set and a second prediction result, and the second prediction result is obtained by simulating on the second operating condition data set. The method further includes determining a corresponding first prediction result by simulating on the first operating condition data set based on the first output result. In this way, a result predicted by the machine learning model can be used as a reference for simulating on operating condition data, thereby improving the accuracy and efficiency of the simulation.