Scholarlink

<p>A thorough understanding of the temperature characteristics of high-speed motorized spindles during operation is crucial for improving machining accuracy. This paper proposes a data-driven temperature prediction method for high-speed motorized spindles, enabling rapid prediction of the temperature field using limited measurement data. First, transient temperature data at different rotational speeds are obtained through finite element simulations. A Kriging-based temperature surrogate model is then developed based on the simulation data. To further improve prediction accuracy, the Northern Goshawk Optimization (NGO) algorithm is employed to optimize the model parameters. The developed model is subsequently used to predict the transient temperature of the motorized spindle, and its reliability is evaluated. Finally, the effectiveness of the proposed model is validated through temperature rise experiments. The results show that the maximum mean absolute error (MAE) between the predicted and experimental values is 0.0811, and the maximum mean squared error (MSE) is 0.069. Compared with finite element simulations, the proposed surrogate model reduces the computation time from 5234&#xa0;s to 28.53&#xa0;s. The proposed method provides an effective solution for temperature prediction of motorized spindles and offers practical value for temperature monitoring, control, and reliability improvement.</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

Temperature prediction of high-speed motorized spindle based on data-driven surrogate model

  • 期刊论文

  • 发表日期：

    2026年5月14日

查看全文
  • Long Shang,
  • Wei Feng,
  • Dongya Li,
  • Baoguo Liu,
  • Yanfang Gao,
  • Yaning Fan,
  • Xinyu Yuan

摘要

A thorough understanding of the temperature characteristics of high-speed motorized spindles during operation is crucial for improving machining accuracy. This paper proposes a data-driven temperature prediction method for high-speed motorized spindles, enabling rapid prediction of the temperature field using limited measurement data. First, transient temperature data at different rotational speeds are obtained through finite element simulations. A Kriging-based temperature surrogate model is then developed based on the simulation data. To further improve prediction accuracy, the Northern Goshawk Optimization (NGO) algorithm is employed to optimize the model parameters. The developed model is subsequently used to predict the transient temperature of the motorized spindle, and its reliability is evaluated. Finally, the effectiveness of the proposed model is validated through temperature rise experiments. The results show that the maximum mean absolute error (MAE) between the predicted and experimental values is 0.0811, and the maximum mean squared error (MSE) is 0.069. Compared with finite element simulations, the proposed surrogate model reduces the computation time from 5234 s to 28.53 s. The proposed method provides an effective solution for temperature prediction of motorized spindles and offers practical value for temperature monitoring, control, and reliability improvement.