Scholarlink

<p>High-voltage power cables manufacturing can be considered continual production system. Optimizing process parameters is significant for continual mass production, especially for large-scale processes. A decision-making agent based on computational intelligence was used to optimize process parameters using signals from phasor measurement units in dynamic processes. This study proposed an innovative framework for optimizing process parameters, specifically for the manufacturing of high-voltage power cables under large-scale processes. A unified benchmark was used to access the optimization decisions of data-driven decision-making agents using the Variational Autoencoder (VAE) model and a Deterministic Policy Gradient (DPG) model. <i>C</i><sub><i>pk</i></sub>, a common process capability metric, was used as an example to evaluate the optimization performance of the VAE and DPG models, and real data from high-voltage cable manufacturing was used to validate the proposed framework. The results showed that the DPG model outperformed the VAE model when the optimal control strategy was unknown. Compared to the existing method, the parameters optimized by the DPG model led to improvements of 24% and 36% in <i>C</i><sub><i>pk</i></sub> for a single production line and a composite production line, respectively.</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

Incorporating process quality capability into process parameter tuning for high-voltage power cable manufacturing

  • 期刊论文

  • 发表日期：

    2025年12月22日

查看全文
  • Chao-Lung Yang,
  • Tzu-Hsien Chu,
  • Guan-Ying Chen,
  • Chae-Wu Huang

摘要

High-voltage power cables manufacturing can be considered continual production system. Optimizing process parameters is significant for continual mass production, especially for large-scale processes. A decision-making agent based on computational intelligence was used to optimize process parameters using signals from phasor measurement units in dynamic processes. This study proposed an innovative framework for optimizing process parameters, specifically for the manufacturing of high-voltage power cables under large-scale processes. A unified benchmark was used to access the optimization decisions of data-driven decision-making agents using the Variational Autoencoder (VAE) model and a Deterministic Policy Gradient (DPG) model. Cpk, a common process capability metric, was used as an example to evaluate the optimization performance of the VAE and DPG models, and real data from high-voltage cable manufacturing was used to validate the proposed framework. The results showed that the DPG model outperformed the VAE model when the optimal control strategy was unknown. Compared to the existing method, the parameters optimized by the DPG model led to improvements of 24% and 36% in Cpk for a single production line and a composite production line, respectively.