摘要
针对航空发动机涡轮盘等曲面金属构件表面裂纹深度高精度检测的迫切需求,提出一种以柔性电磁传感器为基础、脉冲激励与机器学习相结合的定量检测方法,设计差分式柔性电磁传感器与脉冲涡流检测系统,并通过仿真分析得到检测信号多个时频域特征与裂纹深度的关系,在此基础上提出基于人工神经网络的裂纹深度智能反演算法,并通过数据预处理与迁移学习相结合的方法将大量仿真数据和少量实验数据共同应用于反演模型的训练,解决训练样本不足的问题。实验结果表明该方法可实现0~6 mm裂纹深度的定量检测,测量不确定度为0.13 mm,为曲面金属构件的缺陷定量检测提供方法和技术支撑。
In response to the urgent need for high-precision detection of surface crack depth of curved metal components such as aero engine turbine discs,a quantitative detection method based on flexible electromagnetic sensors and combining pulse excitation and machine learning was proposed.A differential flexible electromagnetic sensor and pulse eddy current detection system were designed,and the relationship between multiple time-frequency domain characteristics of the detection signal and crack depth was obtained through simulation analysis.On this basis,an intelligent inversion algorithm for crack depth based on artificial neural networks was proposed.A large amount of simulation data and a small amount of experimental data were applied to the training of inversion model through a combination of data preprocessing and transfer learning,solving the problem of insufficient training samples.The experimental results show that this method can be utilized to detect cracks whose depth ranging from 0 to 6 mm,and the measurement uncertainty of crack depth is 0.13 mm,which providing method and technical support for quantitative detection of defects of curved metal components.
作者
陈棣湘
刘丽辉
任远
曹以恒
陈诗宇
CHEN Dixiang;LIU Lihui;REN Yuan;CAO Yiheng;CHEN Shiyu(College of Intelligence Science and Technology,National University of Defense Technology,Changsha 410073,China;National Key Laboratory of Equipment State Sensing and Smart Support,Changsha 410073,China)
出处
《中国测试》
2025年第9期51-56,共6页
China Measurement & Test
关键词
发动机涡轮盘
裂纹深度
定量检测
柔性传感器
脉冲涡流
engine turbine disc
crack depth
quantitative detection
flexible sensor
pulsed current
作者简介
陈棣湘(1970-),男,湖南湘乡市人,教授,博士生导师,主要从事传感与检测领域的研究。
