Abstract:This paper introduces the application of machine learning methods in industrial informationization, and proposes the construction method of machine learning platform to satisfy the automatic detection requirements in typical equipment production of aviation manufacturing. The method includes: construction of an overall scheme of the machine learning platform, design and optimization of the automatic detection model, training of the detection model and empirical experimental evaluation. This paper takes the automatic detection of diffusion bonding defects of stator vanes as an example, and trains the model with defect detection accuracy which is more than 96%, verifies the feasibility and effectiveness of the machine learning platform.
胡京徽,谢鹏志,杨威,姚罡. 基于扩散连接试样金相照片的非焊合缺陷的自动识别[J]. 航空制造技术, 2020, 63(21): 80-84/97.
HU Jinghui,XIE Pengzhi, YANG Wei, YAO Gang. Automatic Detection of Non-Welding Defects Based on Metallographic Photographs of Diffusion Bonding Samples. Aeronautical Manufacturing Technology, 2020, 63(21): 80-84/97.