基于改进BP神经网络的微裂纹漏磁定量识别
发布时间：2018-12-18 21:25
【摘要】：漏磁检测是铁磁材料常用的无损检测方法之一,定量识别是指通过检测到的漏磁信号识别裂纹的尺寸.采用主成分分析和优化神经网络相结合的建模方法,建立了微裂纹宽度与深度的预测模型.主成分分析去除了数据相关性,减小了输入样本维数,显著简化了网络结构;遗传算法优化的BP神经网络(GA-BP神经网络)可以有效地防止搜索过程中陷入局部最优解.通过基于磁偶极子模型的理论计算与人工刻槽微裂纹漏磁检测实验两种途径验证了该算法在微裂纹定量识别中的应用,为裂纹发展阶段的早期定量识别技术奠定了一定的基础.
[Abstract]:Magnetic leakage detection is one of the commonly used nondestructive testing methods for ferromagnetic materials. Quantitative identification refers to the identification of crack size by detecting magnetic leakage signals. The prediction model of microcrack width and depth is established by combining principal component analysis and optimization neural network. Principal component analysis (PCA) eliminates the data correlation, reduces the dimension of input samples, and simplifies the network structure significantly. The genetic algorithm optimized BP neural network (GA-BP neural network) can effectively prevent the search process from falling into the local optimal solution. The theoretical calculation based on the magnetic dipole model and the experiment of magnetic flux leakage detection of microcracks in artificial grooves verify the application of the algorithm in the quantitative identification of microcracks, which lays a foundation for the early quantitative identification of microcracks in the development stage.
【作者单位】： 北京理工大学机械与车辆学院;机械科学研究总院先进制造技术研究中心;河北环境工程有限公司;
【基金】：国家自然科学基金资助项目(51275048)
【分类号】：TG115.284
，
本文编号：2386522
[Abstract]:Magnetic leakage detection is one of the commonly used nondestructive testing methods for ferromagnetic materials. Quantitative identification refers to the identification of crack size by detecting magnetic leakage signals. The prediction model of microcrack width and depth is established by combining principal component analysis and optimization neural network. Principal component analysis (PCA) eliminates the data correlation, reduces the dimension of input samples, and simplifies the network structure significantly. The genetic algorithm optimized BP neural network (GA-BP neural network) can effectively prevent the search process from falling into the local optimal solution. The theoretical calculation based on the magnetic dipole model and the experiment of magnetic flux leakage detection of microcracks in artificial grooves verify the application of the algorithm in the quantitative identification of microcracks, which lays a foundation for the early quantitative identification of microcracks in the development stage.
【作者单位】： 北京理工大学机械与车辆学院;机械科学研究总院先进制造技术研究中心;河北环境工程有限公司;
【基金】：国家自然科学基金资助项目(51275048)
【分类号】：TG115.284
，
本文编号：2386522
本文链接：https://www.wllwen.com/kejilunwen/jiagonggongyi/2386522.html
