基于BP神经网络管道磁记忆检测模式识别

为区分管道母材及焊缝处不同损伤形式磁记忆信号,运用BP神经网络对管道缺陷检测信号的模式进行识别。以X80管线钢作为试样材质,分别从X80管道母材及焊缝部位取样,加工无缺陷及含裂纹等两种形式的试样并对其进行磁记忆检测。采用有限元分析软件获得其磁场分布,对磁记忆检测信号进行特征参数提取并采用BP神经网络对特征参数进行聚类,建立了管道磁记忆检测模式识别方法。研究结果表明:不同损伤部位及形式的试样,其磁记忆检测信号分布有较大差异;磁记忆检测信号分布与试样表面形貌及损伤形式密切相关;运用BP神经网络能够有效识别管道不同位置及损伤形式的磁记忆检测信号。研究结果为磁记忆检测技术应用于管道内检测并进行管道典型缺陷信号识别提供了新的思路和方法。

Pattern Recognition of Pipeline Magnetic Memory Inspection Based on BP Neural Network

To distinguish the magnetic memory signals of different damage forms at the base material of the pipeline and the weld joint,the BP neural network is used to identify the pattern of the pipeline defect inspection signal.The X80 pipeline steel is used for case study.Samples are taken from the X80 pipeline base material and the weld joint.Two types of samples,defect-free and crack-containing samples,are processed and subjected to magnetic memory testing.Finite element analysis software is used to obtain the magnetic field distribution.The characteristic parameters of the magnetic memory inspection signal are extracted,and are clustered by BP neural network.A pipeline magnetic memory inspection pattern recognition method is established.The research results show that:the samples of different damage locations and types have large differences in the magnetic memory inspection signal distribution.The magnetic memory detection signal distribution is closely related to the sample surface morphology and damage form.The BP neural network can be used to effectively identify pipeline magnetic memory detection signal of different locations and damage forms.The study provides new ideas and methods for the application of magnetic memory inspection technology in inner pipeline inspection and identification of typical pipeline defect signals.