Title: Forecast model for inner corrosion rate of oil pipeline based on PSO-SVM

Authors: Jingcheng Liu; Hongtu Wang; Zhigang Yuan

Addresses: Key Lab for Exploitation of China Southwestern Resources and Environmental Disaster Control Engineering, Ministry of Education, Chongqing University, Chongqing 401331, China. ' Key Lab for Exploitation of China Southwestern Resources and Environmental Disaster Control Engineering, Ministry of Education, Chongqing University, Chongqing 401331, China. ' Key Lab for Exploitation of China Southwestern Resources and Environmental Disaster Control Engineering, Ministry of Education, Chongqing University, Chongqing 401331, China

Abstract: In order to improve the prediction precision of inner corrosion rate of oil pipeline, this paper proposes a novel forecast model, which combines the superior regression performance of a support vector machine and the global optimisation ability of particle swarm optimisation. The SVM regression model, with radial basis function (RBF) kernel, is established to facilitate the inner corrosion rate of oil pipeline and the global optimiser, PSO, is employed to optimise the SVM parameters needed in SVM regression. The proposed model can reduce the dimensionality of data space and preserve features of inner corrosion rate of oil pipeline. The proposed PSO-SVM model, compared with BP neural network model, had higher accuracy and speed and the maximum error is 0.6%. Thus, it provides a new method for the forecast of inner corrosion rate of oil pipeline.

Keywords: inner corrosion rate; oil pipelines; particle swarm optimisation; PSO; support vector machine; SVM; forecast modelling; forecasting; radial basis function; RBF; neural networks; pipeline corrosion.

DOI: 10.1504/IJSPM.2012.047863

International Journal of Simulation and Process Modelling, 2012 Vol.7 No.1/2, pp.74 - 80

Published online: 15 Nov 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article