机械加工
飞秒
材料科学
激光器
非线性系统
涡轮叶片
机械工程
激光打孔
联轴节(管道)
计算机科学
机器学习
人工智能
钻探
支持向量机
机床
高斯分布
刀具
跟踪(教育)
钛合金
空气动力学
非线性规划
非线性回归
水冷
激光加工
制作
算法
涡轮机
螺旋桨
激光切割
数控
热影响区
克里金
作者
Hao Yang,Mingrui Lu,Chongyang Han,Dongyu Han,Dongming Li,boyu sun
标识
DOI:10.1088/1361-6501/ae309b
摘要
Abstract In the manufacture of turbine blade cooling holes for aero-engines, precise control of the heat-affected zone (HAZ) is crucial for ensuring structural integrity, extending service life, and maintaining aerodynamic performance. However, during femtosecond laser drilling of Ti-6Al-4V cooling holes, the HAZ thickness is highly sensitive to processing parameters and exhibits strong nonlinear and multi-factor coupling effects, making its prediction and control particularly challenging. To address this issue, this study proposes a hybrid prediction and optimization approach that integrates machine learning with a genetic algorithm (GA). A Gaussian support vector regression (G-SVR) model is developed to capture the nonlinear relationship between laser parameters and HAZ thickness, while the GA enables multi-objective parameter optimization. The combined G-SVR-GA framework achieves reliable prediction accuracy and provides optimized parameter sets under different objectives, such as minimizing the HAZ thickness or achieving a specified target value. This quantitative and reproducible method offers an effective strategy for HAZ control in femtosecond laser drilling of Ti-6Al-4V cooling holes, providing valuable guidance for precision wall-quality management in the fabrication of critical aero-engine components.
科研通智能强力驱动
Strongly Powered by AbleSci AI