Scholarlink

<p>In the current industrial era, precision is essential and crucial in all aspects of machine tools (MTs); however, MTs are highly susceptible to thermal deformation arising from heat generated by rotating components, electrical subsystems, and environmental temperature changes. Such effects can reduce machining precision, dimensional accuracy, and overall machine performance. Although cooling systems are typically employed to mitigate these effects, varying machining loads and operating conditions require the dynamic adjustment of cooling system parameters. Accurate prediction of thermal deformation is therefore essential to ensure stability and precision in high-performance machining. Therefore, this study proposed an artificial neural network (ANN) model for the thermal deformation prediction of an MT spindle. Since the selection of input parameters plays a critical role in prediction accuracy, a whale optimization algorithm (WOA) was developed to determine the most suitable combinations of cooling system parameters at spindle speeds of 8,000, 10,000, and 12,000&#xa0;rpm. The ANN model demonstrates strong predictive capability with R² values ranging from 0.94 to 0.97 and mean absolute error (MAE) between 0.72 and 0.96 across the tested speeds. The proposed model was compared with ANN-genetic algorithm (ANN-GA) and ANN-firefly algorithm (ANN-FA) models, and the results demonstrated that the proposed ANN-WOA model outperforms the other tested models. To validate the practical feasibility, the optimized cooling parameters were applied in a real CNC machine and experimentally validated in accordance with ISO 230-3 standards, achieving effective control of spindle thermal deformation within 1.12&#xa0;μm. Additionally, the proposed model significantly reduced energy consumption by up to 44.9% and reduced annual CO₂ emissions by approximately 361–480&#xa0;kg, demonstrating its potential for energy-efficient, low-carbon, and sustainable manufacturing.</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

Experimental and machine learning analysis of cooling system optimization for improved thermal stability and energy efficiency in CNC machine tools

  • 期刊论文

  • 发表日期：

    2026年4月14日

查看全文
  • Swami Nath Maurya,
  • Yung-Chuan Chiou,
  • Kun-Ying Li,
  • Abhishek Kumar

摘要

In the current industrial era, precision is essential and crucial in all aspects of machine tools (MTs); however, MTs are highly susceptible to thermal deformation arising from heat generated by rotating components, electrical subsystems, and environmental temperature changes. Such effects can reduce machining precision, dimensional accuracy, and overall machine performance. Although cooling systems are typically employed to mitigate these effects, varying machining loads and operating conditions require the dynamic adjustment of cooling system parameters. Accurate prediction of thermal deformation is therefore essential to ensure stability and precision in high-performance machining. Therefore, this study proposed an artificial neural network (ANN) model for the thermal deformation prediction of an MT spindle. Since the selection of input parameters plays a critical role in prediction accuracy, a whale optimization algorithm (WOA) was developed to determine the most suitable combinations of cooling system parameters at spindle speeds of 8,000, 10,000, and 12,000 rpm. The ANN model demonstrates strong predictive capability with R² values ranging from 0.94 to 0.97 and mean absolute error (MAE) between 0.72 and 0.96 across the tested speeds. The proposed model was compared with ANN-genetic algorithm (ANN-GA) and ANN-firefly algorithm (ANN-FA) models, and the results demonstrated that the proposed ANN-WOA model outperforms the other tested models. To validate the practical feasibility, the optimized cooling parameters were applied in a real CNC machine and experimentally validated in accordance with ISO 230-3 standards, achieving effective control of spindle thermal deformation within 1.12 μm. Additionally, the proposed model significantly reduced energy consumption by up to 44.9% and reduced annual CO₂ emissions by approximately 361–480 kg, demonstrating its potential for energy-efficient, low-carbon, and sustainable manufacturing.