摘要
针对国产大型宽体客机发动机中关键热端承力构件深部三向残余应力场的无损表征难题,开展基于超声测量的残余应力场无损表征与成像方法研究。建立了能够表征任意三向应力影响的声弹性理论模型,揭示了超声体纵波波速相对三向应力变化的定量关系。通过构建含残余应力构件的超声波传播仿真框架,针对典型高温合金锻件设计并采集多角度超声透射数据,分别提出了基于迭代重建算法的层析成像法和基于神经网络的反演成像法。结果表明:层析成像方法能够高灵敏反映锻件内部“内拉外压”分布特征,并对工艺引起的非对称分布具有较高敏感性。神经网络方法则展现出对复杂模式的非线性拟合能力,在高应力区域的平均误差更小。两种方法互为补充,反演结果均与真值保持同一应力水平,能够有效反映实际锻件内部径向、周向、轴向残余应力的幅值与分布状态。所提方法为保障航空发动机关键热端承力构件的尺寸稳定性、可靠性与服役安全性提供了重要数据支撑。
To address the challenge of non-destructively characterizing deep three-dimensional residual stress fields in critical load-bearing hot-section components of domestically developed large wide-body aircraft engines,this study investigates ultrasonic-based methods for non-destructive residual stress characterization and imaging.An ultrasonic acoustoelastic theoretical model capable of describing the influence of arbitrary three-dimensional stresses is established,revealing a quantitative relationship between the relative velocity variations of ultrasonic bulk longitudinal waves and triaxial stresses.By developing an ultrasonic wave-propagation simulation framework for components containing residual stresses,multi-angle ultrasonic transmission data are designed and acquired for representative high-temperature alloy forgings.Two imaging approaches are proposed:a tomographic imaging method based on iterative reconstruction algorithms and an inversion imaging method based on neural networks.The results demonstrate that the tomographic approach can sensitively capture the characteristic“tensile core-compressive surface”stress distribution within the forging and exhibits high sensitivity to process-induced asymmetric distributions.The neural network method,in contrast,shows strong nonlinear fitting capability for complex patterns and yields smaller average errors in high-stress regions.The two methods are complementary,their inversion results remain at the same stress level as the ground truth and effectively reflect the magnitudes and distributions of radial,circumferential,and axial residual stresses within the actual forging.The proposed methodology provides important data support for ensuring the dimensional stability,reliability,and in-service safety of critical load-bearing hot-section components in aero-engines.
作者
李岩锴
袁冰冰
邹睿
卢敬远
王岢
余旭东
邵照宇
邓明晰
LI Yankai;YUAN Bingbing;ZOU Rui;LU Jingyuan;WANG Ke;YU Xudong;SHAO Zhaoyu;DENG Mingxi(School of Astronautics,Beihang University,Beijing 102206,China;State Key Laboratory of High-Efficiency Reusable Aerospace Transportation Technology,Beijing 102206,China;Beijing Key Laboratory of System Design for Reusable Launch Vehicle,Beijing 102206,China;Aero Engine Corporation of China,Commercial Aircraft Engine Co.,Ltd.,Shanghai 200241,China;College of Aerospace Engineering,Chongqing University,Chongqing 400044,China)
出处
《陕西师范大学学报(自然科学版)》
北大核心
2026年第2期19-31,共13页
Journal of Shaanxi Normal University:Natural Science Edition
基金
国家自然科学基金(12134002,12374429,12004026)
中国科协青年人才托举工程项目(2020QNRC002)。
关键词
超声表征
残余应力场
层析成像
神经网络
ultrasonic characterization
residual stress field
tomographic imaging
neural network
