储罐底板漏磁检测量化技术研究  

Research on Signals Quantity Technology of Tank Floor Defects in Magnetic Flux Leakage Testing

作　　者：杨志军[1] 赵海龙[1] 高廷岩[1] 蒋衍彪[1] 

机构地区：[1]大庆石油学院,黑龙江大庆163318

出　　处：《压力容器》2009年第9期7-11,58,共6页Pressure Vessel Technology

摘　　要：储罐是石油、石化工业中重要的设备,储罐底板腐蚀是储罐安全隐患之一。漏磁检测方法是目前储罐底板检测研究的一个重要方向。根据缺陷漏磁信号的特征,将经验模态分解方法(EMD)与小波去噪方法相结合,对漏磁信号进行去噪处理。采用BP神经网络模型对储罐底板缺陷进行量化分析研究,构建了缺陷几何参数预测BP神经网络模型,并运用有限元分析所得到的数据为BP网络训练样本,用人工模拟缺陷的漏磁信号测试BP神经网络。网络训练和测试结果符合储罐底板缺陷量化的精度要求。Tanks are important equipments in petrochemical industry. One of the important reasons which cause the tank to be hazards is tank floor corrosion defects. Magnetic flux leakage testing method is a major direction of tank floor inspection. In this paper, a method is proposed based on wavelet de - noising and empirical mode decomposition （EMD） based on the features of magnetic flux leakage signals. The processing results of measured magnetic flux leakage signals show that good de - noising results can be achieved with this method and the signal noise ratio can be improved. BP neural network model is applied to the quantity analysis of tank floor corrosion defects. A finite element analysis model is established as the training samples. Using the signals of the corrosion defects as the testing samples of the BP neural network. The results in network training and testing reach the quantitative accuracy requirements of tank floor corrosion defects.

关 键 词：储罐底板 漏磁检测 信号处理 量化 

分 类 号：TE972[石油与天然气工程—石油机械设备]