摘要
针对目前合金锯片检测方法效率低、检测缺陷不全面以及误判等问题,提出了一种基于声学共振谱和BP神经网络评分的合金锯片无损检测方法。该方法以共振声学无损检测方法为主体,BP神经网络打分机制辅助来提高检测准确率。首先用模态分析方法分析了不同种类的缺陷对合金锯片声学共振谱的影响,给出了基于物体固有频率差异的声学共振谱检测方法,然后将合金锯片的声学模态信号转成可视化的频域共振谱,最后通过BP神经网络评分评判出每个合金锯片是否满足出厂合格标准。结果表明:该方法拥有精度高、检测缺陷范围广及可靠性好等特点,在锯片无损检测领域有较大的可行性及发展潜力。
In view of carbide saw blade detection method of low efficiency and defect detection are not comprehensive and the miscarriage of justice and other issues,put forward a Nondestructive Testing( NDT) of carbide saw blades based on resonance spectrum and BP neural network score. The resonant acoustic NDT method was taken as the main body,and the BP neural network scoring mechanism was used to improve the detection accuracy. Firstly,using modal analysis method to analyze the influence of different kinds of defects on the carbide saw blade acoustic resonance spectroscopy,acoustic resonance frequency difference object detection method based on spectrum was given,then the acoustic mode signal was turned into a frequency domain signal of carbide saw blade resonance spectrum visualization,and finally through the BP neural network was used to judge whether each blade meet the factory qualified standard. The experimental results show that the method has high accuracy,wide range of defect detection,good reliability and so on,in the field of NDT of saw blade has its feasibility and great potential for development.
作者
王宏民
苏明坤
薛萍
Wang Hongmin;Su Mingkun;Xue Ping(School of Automation,Harbin University of Science and Technology,Harbin 150080,China)
出处
《现代制造工程》
CSCD
北大核心
2019年第8期100-107,共8页
Modern Manufacturing Engineering
关键词
BP神经网络
无损检测
声学共振谱
打分机制
BP neural network
Nondestructive Testing(NDT)
acoustic resonance spectrum
scoring mechanism
