Study On Defects Identification And Judgment Of DC Potential Drop Signature Technology

Petroleum pipeline is one of the important transportation modes of petroleum and natural gas transportation,its construction scale and mileage are continuously developing and improving,but the problem of pipeline safety can not be ignored.Internal wall corrosion is one of the main reasons for pipeline failure,accurate detection and real-time monitoring are very necessary.The DC Voltage Drop Signature Technology is a very effective monitoring technique for buried pipeline,submarine pipeline,high temperature and high pressure equipment,and get more and more recognition at home and abroad.However,the current situation of the technology starting late and being monopolized by foreign companies has greatly restricted its promotion and development in China.Therefore,this paper proposes a study on defect identification and judgment technology of DC Voltage Drop Signature Technology.This subject is combined with finite element simulation and experimental investigation,based on DC Voltage Drop signature Technology,establish the electric field model of the pipeline,design and processing of defective plate samples,use the defect monitoring system device which is developed independently,Measure the potential drop matrix data of the defects,and analyse the causes of current redistribution effect,and study the factors which influence the current redistribution effect,including the defect size,defect location,ambient temperature,current size and precision of data acquisition device,and study the law of influence on electric field distribution.The 2D image processing method is adopted to deal with the potential drop matrix data,realizing the graphical representation of defects,preliminary identify the location and type of defects.Use the SVM&LSSVM to recognize defects and PSO&GA to optimize the parameter.The training and testing of the classification and recognition model of pit erosion and crack defect type are realized,and test and analyse the mearsured data.The GUI will integrate the main contents of this paper,including imaging the potential drop matrix,defect feature data extraction&defect characteristic factor calculation,training the classification and recognition model,verifying the classification model with validation set validation,typing the prediction analysis of test data.