Quantitative Analysis of Circumferential Magnetic Flux Leakage (CMFL) Signal for Oil and Gas Pipeline Based on RBF Neural Network

Abstract

Magnetic Flux Leakage (MFL) testing method has long been used to detect corrosion defect of Oil and Gas pipelines. However, the traditional Axial MFL(AMFL) testing method cannot detect the narrow axial corrosion defect. Applying Circumferential MFL(CMFL) can make up for the deficiency of AMFL. CMFL testing method and its signal analysis is in the initial step in China. Based on ANSYS software, the CMFL model is built to simulate magnetic field. The CMFL signal from simulation model is analyzed with dimensional parameters of corrosion defect, and Radial Basis Function (RBF) neural network method is proposed to quantitatively analyze the relationship between CMFL signal and dimensional feature of corrosion defect, and the detailed procedure of RBF method is given. The results show that MFL signal can qualitatively judge the feature of corrosion defect and RFB neural network method can quantitatively determine the dimension of corrosion defect and provide a reference for Oil and Gas pipeline safety assessment.