Scholarlink

<p>To further enhance the accuracy of thermal error prediction for motorized spindles and establish a compensation model with high predictive precision and strong robustness, this paper proposes an improved whale optimization algorithm (WOA) based on a Cubic chaotic map and adaptive weight adjustment, termed CAWOA. The proposed algorithm is used to optimize a neural network, thereby developing a thermal error prediction model. The temperature field of the motorized spindle was simulated using ANSYS software, identifying the front and rear bearing groups and the motor rotor as the primary heat sources. Experiments were conducted on a VMC0656e five-axis machining center at an ambient temperature of 28&#xa0;°C, where thermal growth and thermal error data were measured at various spindle speeds. A random forest (RF) and recursive feature elimination with cross-validation (RFECV) were employed to select the key temperature measurement points of the spindle, with the filtered temperature rise data used as inputs and the thermal error data as outputs. This data was then used to establish a thermal error prediction model based on a CAWOA optimized BP neural network (CAWOA-BP). The performance of the proposed model was compared against other prediction models, and the results demonstrate that the proposed thermal error compensation model significantly outperforms others in terms of predictive accuracy. This method provides a novel approach for thermal error modeling in motorized spindles.</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

Thermal error prediction of electric spindle based on improved whale algorithm optimized neural network

  • 期刊论文

  • 发表日期：

    2025年6月2日

查看全文
  • Zhongfeng Guo,
  • Siyi Liu,
  • Mingyin Yan,
  • Junlin Yang,
  • Qi Liu,
  • Dongyuan Li

摘要

To further enhance the accuracy of thermal error prediction for motorized spindles and establish a compensation model with high predictive precision and strong robustness, this paper proposes an improved whale optimization algorithm (WOA) based on a Cubic chaotic map and adaptive weight adjustment, termed CAWOA. The proposed algorithm is used to optimize a neural network, thereby developing a thermal error prediction model. The temperature field of the motorized spindle was simulated using ANSYS software, identifying the front and rear bearing groups and the motor rotor as the primary heat sources. Experiments were conducted on a VMC0656e five-axis machining center at an ambient temperature of 28 °C, where thermal growth and thermal error data were measured at various spindle speeds. A random forest (RF) and recursive feature elimination with cross-validation (RFECV) were employed to select the key temperature measurement points of the spindle, with the filtered temperature rise data used as inputs and the thermal error data as outputs. This data was then used to establish a thermal error prediction model based on a CAWOA optimized BP neural network (CAWOA-BP). The performance of the proposed model was compared against other prediction models, and the results demonstrate that the proposed thermal error compensation model significantly outperforms others in terms of predictive accuracy. This method provides a novel approach for thermal error modeling in motorized spindles.