SPH Hyperelastic Simulation Using L-BFGS Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing SPH-based deformable body simulation techniques, such as the 'Fast Corotated Elastic SPH Solids with Implicit Zero-Energy Mode Control', are limited in simulating sophisticated models like neo-Hookean or St. Venant-Kirchoff and suffer from unstable results when using high elastic modulus or large time steps.
Innovation Solution
The method employs a processor and memory to approximate hyperelastic energy using rest-pose volume, material parameters, and deformation gradients, optimizing an objective function through a limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) algorithm to simulate SPH-based deformable bodies, allowing for more accurate and stable simulations with advanced elastic models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the existing backward Euler technique with Cholesky Factorization and Preconditioned Conjugate Gradient is used, then the simulation can handle the simplest Corotated model, but it cannot simulate sophisticated models like neo-Hookean or St. Venant-Kirchoff and produces unstable results with high elastic modulus or large time steps
Solution Approach 1:
The patent transforms the simulation approach by changing the mathematical parameters and solution methodology. Instead of using the traditional backward Euler technique with fixed linear solvers, it introduces a novel objective function formulation combined with L-BFGS optimization, allowing the system to adapt to different hyperelastic models (neo-Hookean, St. Venant-Kirchoff) while maintaining stability through the optimized solution algorithm that properly handles high elastic modulus and large time step conditions
2Device complexity
If the existing linear system solution approach is used, then the computational process is straightforward for simple models, but the simulation accuracy and stability deteriorate when using high elastic modulus or large time steps
Solution Approach 1:
The patent replaces the traditional mechanical linear system solution approach (Cholesky Factorization and Preconditioned Conjugate Gradient) with an optimization-based system using L-BFGS (Limited-memory Broyden-Fletcher-Goldfarb-Shanno) algorithm. This substitution transforms the problem from solving linear systems to minimizing an objective function, which provides better numerical stability and accuracy for hyperelastic simulations with high elastic modulus or large time steps while maintaining computational efficiency through the limited-memory approach
Data Source
AI summary
The present invention discloses a smoothed particle hydrodynamics (SPH)-based hyperelastic simulation method and apparatus. According to the invention, an SPH-based hyperelastic simulation apparatus includes a processor; and a memory connected to the processor, wherein the memory stores program instructions executed by the processor to define states of each of m particles in a discrete time as a set of positions and velocities, for an SPH-based deformable body composed of m particles, approximate hyperelastic energy of each of the m particles using a rest-pose volume, a material parameter, a vectorized deformation gradient, and a projection of the vectorized deformation gradient in order to optimize an objective function for solving new states of each of the m particles, search for an initial approximation of a Hessian matrix using the approximated hyperelastic energy, and simulate the SPH-based deformable body based on the searched initial approximation.