Scholarlink

<p>The high cost of cutting tools, representing approximately 20% of machining expenses, drives the need for strategies to extend tool life and reduce costs. This study integrates deep cryogenic treatment (DCT) with artificial intelligence to optimize face milling parameters for uncoated carbide inserts machining Al7075 alloy. Inserts were treated at − 196&#xa0;°C for 36 h and tempered. Experiments covered various cutting conditions for both treated and untreated tools. To handle data imbalance, the Synthetic Minority Over-sampling Technique for Regression with Gaussian Noise (SMOGN) was applied before training an Extreme Gradient Boosting (XGBoost) model. The model achieved high accuracy, with R<sup>2</sup> values of 0.902 for tool wear and 0.994 for surface roughness. Multi-Objective Particle Swarm Optimization identified optimum parameters: 400 m/min cutting speed, 0.09 mm/rev feed rate, and DCT. Treated tools showed significantly reduced wear. X-ray diffraction revealed a cobalt phase transformation from α-Co (FCC) to ε-Co (HCP), explaining improved wear resistance. Built-up edge was the primary wear mechanism. SHapley Additive exPlanations (SHAP) analysis highlighted cutting speed as dominant for tool wear and feed rate for surface roughness.</p> Graphical abstract <p></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

AI-based optimization of cutting parameters for tool wear and surface roughness of machined specimens

  • 期刊论文

  • 发表日期：

    2025年10月22日

查看全文
  • Attasit Wiangkham,
  • Anantaya Timtong,
  • Metha Kaewta,
  • Amarin Chanpariyavatevong

摘要

The high cost of cutting tools, representing approximately 20% of machining expenses, drives the need for strategies to extend tool life and reduce costs. This study integrates deep cryogenic treatment (DCT) with artificial intelligence to optimize face milling parameters for uncoated carbide inserts machining Al7075 alloy. Inserts were treated at − 196 °C for 36 h and tempered. Experiments covered various cutting conditions for both treated and untreated tools. To handle data imbalance, the Synthetic Minority Over-sampling Technique for Regression with Gaussian Noise (SMOGN) was applied before training an Extreme Gradient Boosting (XGBoost) model. The model achieved high accuracy, with R2 values of 0.902 for tool wear and 0.994 for surface roughness. Multi-Objective Particle Swarm Optimization identified optimum parameters: 400 m/min cutting speed, 0.09 mm/rev feed rate, and DCT. Treated tools showed significantly reduced wear. X-ray diffraction revealed a cobalt phase transformation from α-Co (FCC) to ε-Co (HCP), explaining improved wear resistance. Built-up edge was the primary wear mechanism. SHapley Additive exPlanations (SHAP) analysis highlighted cutting speed as dominant for tool wear and feed rate for surface roughness.

Graphical abstract