Scholarlink

<p>The constant technological evolution of products in the various fields of mechanics has increased competition between designers, forcing them to propose increasingly complex and high-quality designs. The development of software and hardware tools in the CAD/CAM process has made it possible to meet this challenge. However, the development of machining simulation software, which is part of the final stage of this process, has made it possible to virtually interpret the machining process as faithfully as possible in order to predict the quality of the product before it is actually machined. This work deals with the geometric complexity of free-form mechanical parts, which are represented by continuous or discrete models. While continuous models offer accuracy, their implementation is complex. Discrete models are simpler, but their accuracy depends on the discretization density. Taking advantage of advances in software and hardware computer tools, the work focuses on discrete models and proposes a classification approach to intelligently partition free-form parts. This aims to improve the computational efficiency of machining simulations by reducing the processing time. The main contribution of this work is to introduce classification methods “k-means++” and “Mini Batch K-Means” derived from the basic K-means presented in previous works. The results were validated by the development of a graphical and interactive software module that generates a Triple Dexel model and simulates its machining. Tests were carried out on parts of varying complexity. A comparison was made between the classical “cellular” classification method and the proposed K-means techniques (Basic K-Means, K-Means++, and Mini Batch K-Means), which showed performance in terms of computational time compared to the cellular method. With K-Means + + outperforming Basic K-Means. Mini Batch K-Means excelled on parts with high geometric density. Furthermore, the developed approach can be standardized for any type of software dealing with discrete geometric models (mesh).</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

Intelligent free form mechanical parts partitioning: applied in machining simulation

  • 期刊论文

  • 发表日期：

    2025年12月16日

查看全文
  • Khadidja Bouhadja,
  • Younes Abdeldjalil Toumi,
  • Zahida Tchantchane

摘要

The constant technological evolution of products in the various fields of mechanics has increased competition between designers, forcing them to propose increasingly complex and high-quality designs. The development of software and hardware tools in the CAD/CAM process has made it possible to meet this challenge. However, the development of machining simulation software, which is part of the final stage of this process, has made it possible to virtually interpret the machining process as faithfully as possible in order to predict the quality of the product before it is actually machined. This work deals with the geometric complexity of free-form mechanical parts, which are represented by continuous or discrete models. While continuous models offer accuracy, their implementation is complex. Discrete models are simpler, but their accuracy depends on the discretization density. Taking advantage of advances in software and hardware computer tools, the work focuses on discrete models and proposes a classification approach to intelligently partition free-form parts. This aims to improve the computational efficiency of machining simulations by reducing the processing time. The main contribution of this work is to introduce classification methods “k-means++” and “Mini Batch K-Means” derived from the basic K-means presented in previous works. The results were validated by the development of a graphical and interactive software module that generates a Triple Dexel model and simulates its machining. Tests were carried out on parts of varying complexity. A comparison was made between the classical “cellular” classification method and the proposed K-means techniques (Basic K-Means, K-Means++, and Mini Batch K-Means), which showed performance in terms of computational time compared to the cellular method. With K-Means + + outperforming Basic K-Means. Mini Batch K-Means excelled on parts with high geometric density. Furthermore, the developed approach can be standardized for any type of software dealing with discrete geometric models (mesh).