Residual Strength Prediction of Pipeline with Single Defect Based on SVM Algorithm

Abstract

With the continuous development of society, the increasing number and size of oil-gas pipelines, how to ensure the safe operation of the pipeline has become a major problem. At the same time, under the premise of defects, it is of great significance to predict the residual strength of pipeline for the safety of pipeline operation. At present, scholars at home and abroad have conducted in-depth studies on the evaluation of residual strength of pipelines, but there are mainly two problems in the research, namely, low evaluation accuracy and inconvenient application of evaluation method. Therefore, based on SVM algorithm, a residual strength prediction model for single-defect pipelines is established from four aspects, including model construction, parameter optimization, node number determination of hidden layers and kernel function optimization. The prediction results are highly feasible and advanced compared with various algorithms and standard methods.