Research on Failure Probability Prediction Technology of Pressure Pipeline Based on PSO-SVM

Abstract

Pressure pipeline is the key equipment of oil and gas resource transportation. The prediction technology of pipeline failure probability is a global challenge. However, the industrial field data are scattered and have a less amount, which cause a large prediction error, therefore cannot be applied to the actual production. So, this paper makes use of the advantage of support vector machine (SVM) that adapt to small-scale data, and combines with particle filter algorithm (PSO) to find the best SVM parameters, finally builds the probability prediction model of pressure pipeline failure based on PSO-SVM. In order to prove the effect of the proposed model, this paper uses the industrial field data of the pressure pipeline from the oil platform to verify the PSO-SVM model. The results show that the average relative error of the proposed failure probability prediction model is 3.58%, which proves the high accuracy of the PSO-SVM model.