Scholarlink

<p>Stability lobe diagram is essential to determine chatter-free machining parameters for process planning. A tough issue in predicting milling stability is that the crucial inputs tool tip dynamics vary according to tool-holder combination conditions. It is quite time-consuming to update tool tip dynamics through repeated experiments or simulations. To address this issue, this paper proposes a transfer learning and optimization technique-based inverse stability solution to identify the overhang length-dependent tool tip dynamics for different tool-holder combinations. Impact tests are carried out under multiple overhang lengths of a source tool and limited overhang lengths of a target tool. Obtained tool tip FRFs are used to generate analytical stability limits for constructing the source and target datasets respectively. They are combined to iteratively train a regression model for predicting target overhang length-dependent stability limits using instance-based transfer learning, where the domain adaption is implemented by the multi-fidelity surrogate model. Target stability limits under each overhang length are taken to construct an optimization model whose variables are the modal parameters, which is solved by an improved teaching–learning-based optimization algorithm. A detailed experimental study demonstrates that this proposed method can efficiently and accurately predict overhang length-dependent modal parameters requiring fewer impact tests.</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

Transfer Learning and High Dimensional Optimization Supported Tool Tip Dynamics Identification for Different Tool-Holder Assemblies under Inverse Stability Solution

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • Lijun Lin,
  • Chuan Yin,
  • Peng Liu

摘要

Stability lobe diagram is essential to determine chatter-free machining parameters for process planning. A tough issue in predicting milling stability is that the crucial inputs tool tip dynamics vary according to tool-holder combination conditions. It is quite time-consuming to update tool tip dynamics through repeated experiments or simulations. To address this issue, this paper proposes a transfer learning and optimization technique-based inverse stability solution to identify the overhang length-dependent tool tip dynamics for different tool-holder combinations. Impact tests are carried out under multiple overhang lengths of a source tool and limited overhang lengths of a target tool. Obtained tool tip FRFs are used to generate analytical stability limits for constructing the source and target datasets respectively. They are combined to iteratively train a regression model for predicting target overhang length-dependent stability limits using instance-based transfer learning, where the domain adaption is implemented by the multi-fidelity surrogate model. Target stability limits under each overhang length are taken to construct an optimization model whose variables are the modal parameters, which is solved by an improved teaching–learning-based optimization algorithm. A detailed experimental study demonstrates that this proposed method can efficiently and accurately predict overhang length-dependent modal parameters requiring fewer impact tests.