Natural Gas Pipeline Corrosion Rate Prediction Model Based on BP Neural Network

Abstract

AbstractThe transfer medium of oil and natural gas pipelines are flammable, explosive material, and contain various impurities to corrosion, so that internal and external corrosion of pipelines in the conditions are very complex, and pipe’s flaws make the problem more serious. In case of explosion, leakage, shutdown and other accidents, it will lead to serious consequences. In recent years, pipeline spills have occurred, which cause great damage, and are harmful to the environment, so predicting the corrosion rate is very important and meaningful. In this issue, BP neural network is applied to predict the corrosion rate of long distance pipeline. Natural gas pipeline mileage, elevation difference, pipe inclination, pressure, Reynolds number, used as an input parameter and the maximum average corrosion rate of the pipeline used as an output parameter, a prediction model about a natural gas pipeline internal corrosion rate is established. The results show that: BP neural network is of better fitting precision and forecasting result, and prediction of corrosion rate is more reliable based on the model. The results compared with other methods show that, BP neural network algorithm is of fast convergence, with better prediction accuracy, which can predict effectively the corrosion rate of natural gas pipelines and meet the requirements of practical application accurately.KeywordsBP neural networknatural gas pipelineinternal corrosioncorrosion rateprediction model