Skip to content

Turn CAE mesh data => aggregated element feature vectors for ML

License

Notifications You must be signed in to change notification settings

Renumics/mesh2vec

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

98 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Mesh2Vec

Quickstart

mesh2vec

Turn CAE mesh data into aggregated element feature vectors for ML

🚀 Introduction

Mesh2vec is a tool that facilitates the import of Computer-Aided Engineering (CAE) mesh data from LS-DYNA. It utilizes various quality metrics of elements and their surrounding neighborhood to aggregate feature vectors for each element. These feature vectors are of equal length and can be effectively utilized as inputs for machine learning methods. This represents a simpler and more efficient alternative to traditional mesh and graph-based approaches for automatic mesh quality analysis.

⏱️ Quickstart

Installation

  1. Create and activate a virtual environment.
  2. Use the following command to install mesh2vec into your environment:
pip install mesh2vec
  1. Please make sure you have an environment variable ANSA_EXECUTABLE set pointing to your ANSA executable to use ANSA depended features like shell and feature import.
  2. You may temporarily need to set an environment variable SKLEARN_ALLOW_DEPRECATED_SKLEARN_PACKAGE_INSTALL=True.

Load Mesh

from pathlib import Path
from mesh2vec.mesh2vec_cae import Mesh2VecCae
m2v = Mesh2VecCae.from_ansa_shell(4,
    Path("data/hat/Hatprofile.k"),
    json_mesh_file=Path("data/hat/cached_hat_key.json"))

Add element features

m2v.add_features_from_ansa(
    ["aspect", "warpage"],
    Path("data/hat/Hatprofile.k"),
    json_mesh_file=Path("data/hat/cached_hat_key.json"))

Aggregate

import numpy as np
m2v.aggregate("aspect", [0,2,3], np.nanmean)

Extract Feature Vectors

m2v.to_dataframe()

data frame with feature vectors

Optional: Visualize a single aggregated feature on mesh

m2v.get_visualization_plotly("aspect-nanmean-2")

3d mesh plot of agggredated