Quantitative Technology And Application Research On Magnetic Flux Leakage Inspection Of Pipeline Defects

This dissertation concentrates on the difficult problems that the pipeline or pipe defects are not estimated or evaluated quantitatively, intelligently by MFL (Magnetic Flux Leakage) inspection method. From the practical demand, theoretical analysis and experiments or testing, the MFL inspection technology of pipeline or pipe defects are explained, summarized in detail in the whole paper, at the same time, the relationship between the MFL field distribution, the MFL signal shape and the defect geometry or severity, the analysis of the MFL data and the compensation of influencing the relationship factors, signal feature extraction, intelligent recognition of defect parameters and so on are studied systematically. The main achievements are as follows:The theory models of defect MFL field are set up; based on the magnetic dipole model and finite element model, the distribution of defect MFL field is simulated in the dissertation; the component of flux density (magnetic filed) parallel to the surface of pipe is detected by a circumferential array of Hall probes; the factors including defect geometry parameters, defect surface angle and shape, pipe material, field intensity, vehicle velocity, background magnetization, operating pressure and remanent magnetization and so on influence the MFL signal characteristics. The signals are pre-processed and transformed by each sensor gain amplification and accordance, signals differential conversion, digital filter and smooth; spatial signals are converted into time domain signals; from the time sampling data, the processing of time-domain, frequency-domain, time-frequency domain analysis, the elimination of signal noise and disturbance, are implemented in this dissertation. The nonlinear interpolation method is used to compensate the influence of pipe grade and pipe material property; tool velocity effects are compensated by the Fourier transformation and optimal filter theory. The pattern recognition method of pipe MFL signals is put forward, the features of signals are extracted from the recorded flux leakage response and characterizing definition is introduced as well; the main-part analysis, nonlinear regression, statistical methods are studied and used to establish characterization and compensation algorithms, the quantitative estimation of defect geometry and the result accuracy are accepted. Artifical neural networks are employed for defect accurate recognition and calculation, the traditional BP neural network and wavelet basis function neural networks can successfully predict or estimate defect shape and geometry parameters, their application to the magnetic flux leakage inspection is put forward at first. The intelligent inspection system of pipeline is established and the project database is designed, the interpolation and signature map of defect MFL signal data are solved effectively, the feature automatic extraction and intelligent evaluation of geometric parameters of the defects are shown and proved practical; the whole process of pipeline inspection are given step by step, these results and conclusions help improve the pipe defect recognition and solve some difficult problems in the field of pipeline MFL inspection.