节点文献
基于机器学习的硬质合金微钻孔位精度预测
Machine Learning-Based Prediction of Hole Position Precision in Cemented Carbide Micro-Drills
【摘要】 硬质合金微钻的孔位精度是评价印制电路板钻孔加工质量的核心指标,与材料成分、微观组织及力学性能的协同作用紧密关联。本研究聚焦0.15 mm钻径的硬质合金微钻,构建了系列机器学习模型以量化其成分、晶粒尺寸、力学性能与钻孔精度的关系,其中梯度提升回归(GBR)模型在预测性能和钻孔精度上最为出色。通过对影响因素的分析可知,在力学性能指标中,抗弯强度对孔位精度的贡献占比最高(43%),显著高于硬度(25%)和断裂韧性(32%);在成分与组织特征中,Co含量对综合力学性能的影响权重最大,其次为WC晶粒尺寸,而Cr3C2和VC含量的影响较小。借助所建机器学习模型,预测得到兼具高综合力学性能与高孔位精度的硬质合金微钻的典型成分及晶粒尺寸参数。本研究证实了数据驱动方法在微钻性能优化中的实用价值,为硬质合金微钻的成分与组织设计提供了定量参考,可直接用于指导高性能、高精度微钻的制备,助力集成电路领域关键材料制备技术的升级。
【Abstract】 The hole position precision of cemented carbide micro-drills is a critical indicator for evaluating the quality of printed circuit board drilling, which is closely associated with the synergistic effects of material composition,microstructure, and mechanical properties. This study focused on cemented carbide micro-drills with a diameter of 0.15 mm and developed a series of machine learning models to quantify the relationships among their composition, grain size,mechanical properties, and drilling precision. Among these models, the gradient boosting regression(GBR) model demonstrates superior performance in predicting both mechanical properties and drilling precision. Analysis of influencing factors regarding the mechanical properties reveals that transverse rupture strength contributes the most to hole position precision(43%), significantly higher than hardness(25%) and fracture toughness(32%). Regarding compositional and microstructural characteristics, Co content has the greatest impact on comprehensive mechanical properties, followed by WC grain size, while the influences of Cr3 C2 and VC contents are relatively minor. By using the established machine learning model, the typical composition and grain size parameters for cemented carbide micro-drills with both high comprehensive mechanical properties and high hole position accuracy are predicted. This study confirms the practical value of data-driven methods in optimizing micro-drill performance and provides quantitative guidance for the composition and microstructure design of cemented carbide micro-drills, which can be directly applied to guide the fabrication of highperformance and high-precision micro-drills, thereby supporting advancements in key material preparation technologies for the integrated circuit industry.
【Key words】 cemented carbide micro-drill; machine learning; composition and grain size; mechanical property; hole position precision;
- 【文献出处】 硬质合金 ,Cemented Carbides , 编辑部邮箱 ,2025年06期
- 【分类号】TG52
- 【下载频次】8