基于GA-BP神经网络的微裂纹漏磁检测定量化评价  

Evaluation on Quantitative Recognition of Micro Cracks by Magnetic Leakage Test Based on GA-BP Neural Network

作　　者：邱忠超[1] 张卫民[1] 果艳 刘金[2] 成明明[2] 

机构地区：[1]北京理工大学机械与车辆学院,北京100081 [2]北京理工大学宇航学院,北京100081

出　　处：《无损检测》2016年第2期1-4,共4页Nondestructive Testing

摘　　要：介绍了利用遗传算法优化BP神经网络,实现金属中微细裂纹漏磁检测定量化评价的基本原理。将遗传算法和人工神经网络有机结合,进行漏磁定量化检测,既提高了算法的全局搜索性,又良好地适应于非线性问题。试验结果表明,将该人工智能算法应用于工程实际,能有效实现基于漏磁检测信号的金属中微细裂纹定量化评价。The basic principle of realizing quantitative evaluation of metal micro crack detection with using BP neural network optimized by genetic algorithm is introduced.Organic combination of genetic algorithm and artificial neural network not only improves global search performance,but also maintains good adaptability to nonlinear problems during magnetic flux leakage detection.Final experimental results show that the artificial intelligence algorithm applied in practical engineering can realize quantitative assessment of metal micro cracks based on magnetic leakage signals.

关 键 词：GA-BP神经网络 微裂纹 漏磁检测 定量化 

分 类 号：TG115.28[金属学及工艺—物理冶金]