Machine-Learned PLIC for Faster Interface Construction on Mesh Cells
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing piecewise linear interface construction (PLIC) methods in numerical simulations, such as those used in Volume of Fluid (VOF) algorithms, are computationally slow due to complex geometrical operations and iterative or analytical solutions, especially for irregular mesh shapes.
Innovation Solution
A machine learning approach using a neural network algorithm trained on normalized geometry information of various mesh types to predict piecewise linear interfaces, reducing the need for iterative calculations and improving computation efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional iterative or analytical methods are used to solve PLIC problems, then accurate interface construction is achieved, but computation time increases significantly
Solution Approach 1:
The patent pre-computes and stores interface construction solutions for various cell fractions and normal vectors in lookup tables during an offline training phase. During actual simulations, the pre-computed solutions are directly retrieved based on the current cell fraction and normal vector, eliminating the need for time-consuming iterative or analytical calculations at runtime while maintaining accuracy.
Solution Approach 2:
The patent creates simplified representations of the complex geometrical relationships by training machine learning models (neural networks) on synthetic data generated from analytical solutions. The trained models copy the essential mapping between input parameters (cell fraction, normal vector) and output interfaces, enabling fast prediction without repeating the complex geometrical operations.
2Manufacturing precision
If complex geometrical operations are performed to construct interfaces, then accurate piecewise linear interfaces are obtained, but computational complexity increases
Solution Approach 1:
The patent replaces complex geometrical calculation systems with machine learning prediction systems. Instead of performing iterative geometrical operations to construct interfaces, the system uses trained neural networks that have learned the geometric relationships during training. This substitution eliminates complex runtime geometrical operations while preserving interface construction accuracy.
Solution Approach 2:
The patent transforms the PLIC problem from a geometrical construction problem into a parameter prediction problem. By changing the approach from solving geometrical equations to predicting interface parameters directly from cell fraction and normal vector inputs, the computational complexity is significantly reduced while maintaining precision.
3Ease of manufacture
If analytical solutions with iterative steps are used for simple geometric shapes, then PLIC problems are solved, but computation speed decreases
Solution Approach 1:
The patent extracts the computationally expensive iterative solving step from the overall PLIC process and replaces it with direct lookup or machine learning prediction. The essential functionality of finding the interface position is separated from the complex iterative procedure, allowing the remaining steps to be executed efficiently without the bottleneck of iterative convergence.
Data Source
AI summary
Methods, systems, and apparatus, including medium-encoded computer program products, for a machine learning approach for piecewise linear interface construction include: obtaining a cell fraction for a mesh cell in a mesh, a normal vector perpendicular to a linear interface for the mesh cell, and geometry information of the mesh cell, wherein the geometry information includes at least two data values; normalizing the geometry information of the mesh cell to reduce the at least two data values to at least one data value; obtaining an output value from a machine learning algorithm, previously trained using normalized geometry information of multiple different cells of a same mesh type as the mesh cell; determining the linear interface for the mesh cell based on the output value, the normal vector and the geometry information; using the linear interface for the mesh cell in numerical simulation processing of the mesh.