Scholarlink

<p>The fundamental goal of materials science and engineering is the design of better materials and this is done through extensive experimental and numerical investigations, which become more complex as the field progresses. The integration of machine learning methods into this process can potentially revolutionize the field, but requires large datasets. A new method, called <i>synthetic dataset generation</i> promises to reduce this effort by using validated numerical simulations as inputs for machine learning. However, this method requires tools for creation and analysis of numerical models. In this paper, we introduce the free and open-source <i>VCAMS</i> software package which can be used to create very complex, multi-material finite element models with great accuracy, speed, and reproducibility. The software has a graphical user interface (GUI) for convenient use, and has a scripting interface for power users. It also can readily interface with most scientific and machine learning libraries in the Python programming language. This paper first introduces the background, concepts, and contents of the software package and then provides three case studies to showcase the capabilities of the software in different fields: The first case study demonstrates creation of a micromechanical model based on a microscopy image. The second case study models a 3D composite with spherical inclusions and periodic boundary conditions. And the final case study introduces a <i>data generation pipeline</i> that automatically creates, prepares, analyzes, and post-processes 362 micromechanical models. The software is actively developed and used by our group and its source code and documentation are available online along with and all of the data and scripts used in the paper.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A free and open-source software for fast and reproducible voxel-based modeling of complex structures

  • 期刊论文

  • 发表日期：

    2026年4月28日

查看全文
  • Mohammadreza Khoshbin,
  • Ali Pourkamali Anaraki,
  • Ali Cheloee Darabi

摘要

The fundamental goal of materials science and engineering is the design of better materials and this is done through extensive experimental and numerical investigations, which become more complex as the field progresses. The integration of machine learning methods into this process can potentially revolutionize the field, but requires large datasets. A new method, called synthetic dataset generation promises to reduce this effort by using validated numerical simulations as inputs for machine learning. However, this method requires tools for creation and analysis of numerical models. In this paper, we introduce the free and open-source VCAMS software package which can be used to create very complex, multi-material finite element models with great accuracy, speed, and reproducibility. The software has a graphical user interface (GUI) for convenient use, and has a scripting interface for power users. It also can readily interface with most scientific and machine learning libraries in the Python programming language. This paper first introduces the background, concepts, and contents of the software package and then provides three case studies to showcase the capabilities of the software in different fields: The first case study demonstrates creation of a micromechanical model based on a microscopy image. The second case study models a 3D composite with spherical inclusions and periodic boundary conditions. And the final case study introduces a data generation pipeline that automatically creates, prepares, analyzes, and post-processes 362 micromechanical models. The software is actively developed and used by our group and its source code and documentation are available online along with and all of the data and scripts used in the paper.