Scholarlink

<p>Machining-induced residual stress critically governs the in-service fatigue life and reliability of thin-walled parts. Current research predominantly focuses on depth-wise residual stress, whereas investigations into the surface residual stress (SRS) field and its inherent anisotropy remain insufficient. This gap stems from the fact that SRS constitutes a tensor field, characterized by bidirectional coupling between orthogonal components at each material point, which conventional scalar or analytical predictive models fail to capture. Thus, this paper presents a novel physics-constrained prediction method for the SRS fields based on the spatial distribution of effective milling power. Firstly, a kinematic model is developed to calculate the effective milling power by integrating the five-axis toolpath with individual cutting-edge motion. Then, the spatial distribution of effective milling power in multi-pass milling is modeled by accounting for energy time-lag dissipation. Finally, a physics-constrained framework maps the power distribution to biaxial SRS fields while explicitly enforcing mechanical equilibrium and energy conservation. Experimental validation demonstrates the performance of proposed model, reducing RMSE and MAE by 32.8–45.7% and elevating R2 from 0.61 to over 0.87. Moreover, the model generates physically credible stress fields by enforcing mechanical equilibrium and energy conservation, evidenced by markedly reduced gradient direction similarity. The proposed method bridges manufacturing physics with data‑driven learning, providing a novel framework for the digital characterization and precision control of machining‑induced residual stress.</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

A novel physics-constrained prediction method for bidirectional surface residual stress fields based on the spatial distribution of effective milling power

  • 期刊论文

  • 发表日期：

    2026年7月14日

查看全文
  • Zehua Wang,
  • Changjiu Xia,
  • Guofu Ding,
  • Lei Jiang,
  • Xin He,
  • Sibao Wang

摘要

Machining-induced residual stress critically governs the in-service fatigue life and reliability of thin-walled parts. Current research predominantly focuses on depth-wise residual stress, whereas investigations into the surface residual stress (SRS) field and its inherent anisotropy remain insufficient. This gap stems from the fact that SRS constitutes a tensor field, characterized by bidirectional coupling between orthogonal components at each material point, which conventional scalar or analytical predictive models fail to capture. Thus, this paper presents a novel physics-constrained prediction method for the SRS fields based on the spatial distribution of effective milling power. Firstly, a kinematic model is developed to calculate the effective milling power by integrating the five-axis toolpath with individual cutting-edge motion. Then, the spatial distribution of effective milling power in multi-pass milling is modeled by accounting for energy time-lag dissipation. Finally, a physics-constrained framework maps the power distribution to biaxial SRS fields while explicitly enforcing mechanical equilibrium and energy conservation. Experimental validation demonstrates the performance of proposed model, reducing RMSE and MAE by 32.8–45.7% and elevating R2 from 0.61 to over 0.87. Moreover, the model generates physically credible stress fields by enforcing mechanical equilibrium and energy conservation, evidenced by markedly reduced gradient direction similarity. The proposed method bridges manufacturing physics with data‑driven learning, providing a novel framework for the digital characterization and precision control of machining‑induced residual stress.