Scholarlink

<p>Chatter is a common issue during machining operations and significantly affects machining quality. A stability lobe diagram (SLD) is typically used to mitigate chatter. However, the generated SLD needs frequent adjustments due to its susceptibility to various time-varying factors, such as tool wear. In this study, we propose a novel method for generating an SLD that maintains accuracy despite the influence of various time-varying factors. The SLD is generated using a learning digital twin (LDT). We begin by proposing the structure of the LDT. Two concept drift-based algorithms are introduced to detect drifts in the behavior of the physical machine. Additionally, a new hybrid model structure is proposed to simulate the machine’s behavior. The modified SLD is subsequently generated through the evaluation of the simulated behavior from the hybrid model. The results demonstrate that the changes in the physical machine can be accurately detected, and the modifications to the SLD achieve high accuracy. We conclude that the LDT can be used for generating the SLD and maintains high accuracy throughout the lifetime of the physical machine.</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

Learning digital twin: a case study on chatter suppression based on a time-varying stability lobe diagram

  • 期刊论文

  • 发表日期：

    2025年3月5日

查看全文
  • Zhibo Zhou,
  • Michael Walther,
  • Alexander Verl

摘要

Chatter is a common issue during machining operations and significantly affects machining quality. A stability lobe diagram (SLD) is typically used to mitigate chatter. However, the generated SLD needs frequent adjustments due to its susceptibility to various time-varying factors, such as tool wear. In this study, we propose a novel method for generating an SLD that maintains accuracy despite the influence of various time-varying factors. The SLD is generated using a learning digital twin (LDT). We begin by proposing the structure of the LDT. Two concept drift-based algorithms are introduced to detect drifts in the behavior of the physical machine. Additionally, a new hybrid model structure is proposed to simulate the machine’s behavior. The modified SLD is subsequently generated through the evaluation of the simulated behavior from the hybrid model. The results demonstrate that the changes in the physical machine can be accurately detected, and the modifications to the SLD achieve high accuracy. We conclude that the LDT can be used for generating the SLD and maintains high accuracy throughout the lifetime of the physical machine.