Identification of Tank Bottom Plate Defects Based on LSTM
Leakage caused by corrosion perforation or fatigue cracks on the bottom plate of storage tanks is the main failure form of atmospheric storage tanks.The detection of tank bottom plates usually adopts electromagnetic technology,which utilizes the leakage magnetic effect generated by permanent magnet magnetization of steel plates for defect detection.However,this detection method cannot distinguish between crack defects and corrosion defects in steel plates,which brings inconvenience to later processing.In this paper,the magnetic testing signals including crack defects and corrosion defects are used as data sources,and a one-dimensional time series neural network model based on LSTM(Long short-term memory)is established to realize the classification and recognition of tank floor defects.The results show that the neural network model based on LSTM can quickly classify and identify tank bottom plate defects,with an accuracy of 96%for i-dentifying steel plate crack defects and 92%for identifying steel plate corrosion defects.
tank bottom platemagnetic flux leakage testingdefect identificationLSTM
万方数据
维普
