The invention relates to the technical field of pipeline detection, in particular to an intelligent non-contact pipeline detection
system based on a
metal magnetic memory effect, and provides the following scheme: dividing a to-be-detected pipeline into a plurality of grid units, obtaining self-leakage
magnetic field data of each grid, constructing a
magnetic moment matrix, and forming a
magnetic field matrix; according to the method, abnormal grids are positioned based on a
magnetic field matrix, a grid with the highest abnormal
score is selected from a spatially continuous abnormal set through screening logic to serve as a seed grid, authenticity of adjacent abnormal grids is verified step by step in combination with a local inversion window and
magnetic flux continuity constraint, false anomalies caused by
magnetic flux diffusion or
noise are eliminated, and the accuracy of the abnormal grids is improved. Performing
magnetic moment matrix updating on the confirmed real defect grid; and finally, calculating a
magnetic gradient modulus by utilizing the updated
magnetic moment matrix and determining a damage grade. Adjacent lattice infection can be effectively inhibited, the defect positioning precision and the damage evaluation reliability are improved, and the method is suitable for a non-contact inspection scene of an in-service pipeline.