Quantitative Inversion of Pipeline Defect Depth Based on Support Vector Machine and Magnetic Memory Technology

Various depths of corrosion defects often occurred on the surfaces of metal pipelines. At present metal magnetic memory detection technology is the only non-destructive testing technology that can diagnose the early damage of ferromagnetic components. However, the original signal of magnetic memory cannot directly realize the quantitative identification of pipeline corrosion defects, and thus it is impossible to realize the warning of the degree of corrosion of the pipeline. Aiming at this problem, a quantitative inversion model of pipeline defect depth was established using support vector machine method. The model was used to identify and predict the corrosion defects with a depth of 1-15 mm on the metal pipelines. The average error of the prediction results was 2.398 mm, and the average root mean square error was 3.205 mm. The results demonstrate that the model was feasible for quantitative inversion of pipeline corrosion depth. The research results could provide a certain reference for the research in this field, and had high practical application value.