A General Inversion Method Based on Magnetic Flux Leakage Inspection

Aiming at the problem that the transfer accuracy of profile inversion model in different pipeline fields is not high enough in magnetic flux leakage (MFL) inspection, this paper presents a general inversion method. Firstly, the defect features of different pipelines are transferred by transfer component analysis (TCA), to reduce the difference of data probability distribution between them. Then, the post-transfer defect features are used to perform defect inversion with random forest (RF) algorithm. The real data are required from the domestic in-service oil pipelines in experiments. The experimental results show that the proposed method can effectively develop inversion accuracy by applying the inversion model into other pipeline fields.