Research and Application of a Corrosion Prediction Method Based on Internal Detection Database

Abstract

Gas gathering and transmission pipelines are often located in high corrosion risk operating environments, which are prone to metal corrosion, perforation, and cause safety accidents and economic losses. Scientifically and reasonably predicting the corrosion rate of pipelines is an effective means to avoid corrosion perforation accidents. Therefore, a corrosion prediction method based on an internal detection database has been developed. This method is based on a self-built internal detection database for gas gathering pipelines, and the prediction of pipeline corrosion rate is achieved by establishing a wavelet neural network (WNN) model optimized by genetic algorithm (GA). The application results of the example show that the proposed GA-WNN corrosion rate prediction model has an average absolute error of 0.0106mm/a and an average relative error of 10.99%, with high accuracy. It can be used as a good tool for predicting the corrosion rate of gas gathering and transportation pipelines.