Internal and external defect identification of pipelines using the PSO-SVM method Defects due to corrosion can occur on both the internal and external surfaces of a pipeline. Magnetic flux leakage (MFL) pipeline inspection gauges can detect and locate defects on both the internal and external wall surfaces of pipelines but are generally unable to discriminate between internal and external defects. Therefore, this paper presents a classification approach to achieve defect discrimination based on support vector machines (SVM). First, time-frequency analysis is employed to acquire the feature vectors of the time domain and frequency domain from the MFL signal. Then, a distance evaluation technique is used to remove the redundant and irrelevant information and select the salient features for the classification process. Next, the selected feature vectors are used as the inputs to the proposed SVM model of defect classification. The parameters of this model are optimised using the particle swarm optimisation (PSO) algorithm. Finally, with the optimised model parameters and training sample data, the classification model of the location of the defect based on SVM is established. The experimental results indicate that the proposed method is an effective approach for internal and external defect classification based on the MFL signal.

No abstract