Scholarlink

<p>Thermal errors significantly impact the precision of computer numerical control (CNC) machine tools, necessitating effective compensation strategies. This study proposes a novel approach for temperature-sensitive point selection and thermal error modeling in CNC lathing machines. The methodology integrates thermal imaging for heat source identification, statistical analysis through Independent Samples T-test for temperature point screening, and Grey Relation Analysis for determining axis-specific temperature-sensitive points. Two neural network architectures—Feedforward Neural Network and Non-linear Input-Output Neural Network Time-Series (NIONNTS)—were developed and compared for thermal error prediction. The NIONNTS model demonstrated superior performance, with its stability validated using untrained data. Experimental results showed significant improvements in thermal error compensation: the Z-axis error reduced from 39.8 to 2.5&#xa0;μm, the Y-axis from − 6.4 to 1.6&#xa0;μm, and the X-axis from − 5.0 to − 4.4&#xa0;μm. These results validate the effectiveness of the proposed methodology in enhancing machine tool precision through thermal error compensation<b>.</b></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 Novel Approach for Temperature-Sensitive Point Selection and Thermal Error Modeling in CNC Lathing

  • 期刊论文

  • 发表日期：

    2025年7月31日

查看全文
  • Wen-Lin Chu,
  • Jun-Ye Luo,
  • Bo-Lin Jian

摘要

Thermal errors significantly impact the precision of computer numerical control (CNC) machine tools, necessitating effective compensation strategies. This study proposes a novel approach for temperature-sensitive point selection and thermal error modeling in CNC lathing machines. The methodology integrates thermal imaging for heat source identification, statistical analysis through Independent Samples T-test for temperature point screening, and Grey Relation Analysis for determining axis-specific temperature-sensitive points. Two neural network architectures—Feedforward Neural Network and Non-linear Input-Output Neural Network Time-Series (NIONNTS)—were developed and compared for thermal error prediction. The NIONNTS model demonstrated superior performance, with its stability validated using untrained data. Experimental results showed significant improvements in thermal error compensation: the Z-axis error reduced from 39.8 to 2.5 μm, the Y-axis from − 6.4 to 1.6 μm, and the X-axis from − 5.0 to − 4.4 μm. These results validate the effectiveness of the proposed methodology in enhancing machine tool precision through thermal error compensation.