基于超声波检测的BP神经网络缺陷识别方法设计

针对石油储罐底板的缺陷损伤问题,建立局部储罐底板模型,应用超声波缺陷检测技术进行分析研究。利用ABAQUS有限元软件进行超声检测的模拟仿真分析,采集并保存接收信号,通过不同位置的发射/接收器信号时间差值对比,应用椭圆定位原理计算得到缺陷类型;从仿真模拟数据中选取不同的超声波回波信号作为神经网络的输入,借助于MATLAB软件训练出可识别底板缺陷的BP神经网络,选取多组数据进行测试,经验证,设计的BP神经网络对缺陷数据识别具有一定正确性。

Design of BP Neural Network Defect Recognition Method Based on Ultrasound Detection

For the defect damage problem of oil tank baseplate,a local tank baseplate model was established,and the ultrasonic defect detection technology was applied to analyze the defect. The ABAQUS finite element software was used to simulate and analyze the ultrasonic detection,the received signal was collected and saved. By comparing the time difference between the transmitting and receiving signals at different locations,the defect type was obtained through calculation by using ellipse positioning principle. Different ultrasonic echo signals were selected from the simulation data as the input of the neural network,and the BP neural network of the identifiable baseplate defect was trained with the aid of the MATLAB software,and was tested by selecting multiple sets of data. It was verified that the designed BP neural network can correctly identify defect data to certain degree.