Surrogate Model Validation for Simulation Fidelity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Developing simulation models with high fidelity and accuracy is time-consuming and requires extensive physical testing, making it challenging to reduce the effort and time needed for creating simulation models for physical structures.
Innovation Solution
A computer system trains a machine learning model to generate surrogate models based on physical test data, selects simulation values using a cost function, and compares simulation results with physical test results to iteratively improve the model's accuracy, reducing the reliance on physical testing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If extensive physical testing is performed to develop simulation models, then model fidelity and accuracy are improved, but time consumption and effort increase
Solution Approach 1:
The system performs preliminary physical testing to collect training data, then uses this data to train a surrogate model that can predict simulation outcomes without requiring additional physical testing. This preliminary action establishes a foundation that reduces subsequent time requirements while maintaining model fidelity.
Solution Approach 2:
The patent creates a surrogate model that copies the behavior of the complex physics-based simulation model. This surrogate model serves as a simplified copy that can be trained on physical test data and used for predictions without requiring extensive new physical testing, thus reducing time consumption while preserving accuracy.
2Manufacturing precision
If iterative training with multiple physical tests is performed, then simulation model accuracy is improved, but the number of physical tests and effort required increases
Solution Approach 1:
The system implements feedback by comparing surrogate model predictions with actual physical test results, then using this comparison to iteratively retrain and improve the surrogate model. This feedback loop continues until the model achieves sufficient accuracy, reducing the need for extensive upfront physical testing while maintaining high simulation accuracy.
Solution Approach 2:
The patent applies partial action by performing physical testing only to the extent necessary to train the surrogate model adequately. Rather than conducting exhaustive physical tests, the system uses a subset of physical test data to create a surrogate model that can then handle the majority of prediction tasks, improving productivity while maintaining acceptable accuracy levels.
Data Source
AI summary
A method, apparatus, system, and computer program product for managing a physics simulation model. A machine learning model is trained to output predicted test results for sets of simulation values for a set of simulation parameters using a training data set based on test results for physical structures to form a surrogate model. Current simulation values for simulation parameters are selected using the surrogate model and a cost function. Simulation test results are generated using the physics simulation model that implements the current simulation values selected for the simulation parameters. The simulation test results are compared with physical test results from testing the set of physical structures using physical test inputs applied to the physical structures to form a comparison. The surrogate model is trained using the current simulation values selected for the simulation parameters using the surrogate model in response to the comparison being outside of a tolerance.