Research On The Analysis Method Of The Magnetic Flux Leakage Testing Signal For Pipeline Defects

MFL(Magnet Flux Leakage) testing is the common way and one of the most effective way of the nondestructive detection,which plays an very important role on protecting the oil field's running pipings safety.Compared to other detective ways,it has many advantages as follows:high detection sensitivity,high reliability,quantizing defects, operating Easily,suitable for large-area detection.It also has the prominent detection effect,quick speed detection,high efficiency on inspecting the defect of eccentric wear and corrosion pits and other forms.However,as compared to MFL inspection equipment development,the existing MFL defect assessment mainly depends on the personnel experiences,which leads to large workload,low efficiency,large subjective factors influence.But between the defect geometry parameters and magnetic flux leakage signal, it has non-linear characteristics.And the magnetic flux leakage coefficient of pipe material differences and many other factors influencing MFL signals increase the difficulties of defect analysis.So,with the purpose of solving the many existing problems,this paper uses the finite element simulation software ANSYS and the RBF(radial basis function) neural network method to study the magnetic flux leakage signal of crack defects and have a quantitative analysis of defects through the MFL signal.The main contents are as follows:First,the MFL technical principle,producing,development and application have been introduced.It also introduces the defect theory calculation of MFL analysis and the MFL detecting device principle,the working process,the role of the main components, working principle,design methods and the ways of working.Secondly,after the magnetization mode and designing the part size have been chosen, convert the actual measured target into two-dimensional geometric model and use the finite element simulation software ANSYS to simulate the crack defects.By defining of an optimal path,we can draw the radial and axial magnetic flux density curve of MFL. According to the characteristic quantities of curves,we can obtain the law relationship between the crack of the defect characteristics and magnetic flux leakage signal.Finally,set up the defect data sample library by the analysis data.Select the 45 characteristic quantities related the defects' geometric parameters as the training vector from the sample database,defects' geometric parameters as the target vector,and the remaining 5 groups as the detected target.And then use of RBF neural networks for quantitative analysis of defects.Neural network prediction results can be shown that this method can be used for quantitative identification of crack defects.