An Efficient Corrosion Prediction Model Based on Genetic Feedback Propagation Neural Network

Abstract

Corrosion is one of the most significant challenges for oil pipelines. It can occur due to various factors such as moisture, oxygen, and contaminants in the oil. Corrosion weakens the pipeline material, leading to leaks, ruptures, and structural failure. To enhance the ability to decrease the corrosion problems of oil pipelines, an efficient Back Propagation Neural Network is developed to predict the corrosion rate and analyse the importance of the features that affect the corrosion. This method is based on the database generated by coupling an analytical corrosion rate model and Monte Carlo simulation by using Spearman’s (SP) correlation coefficient to generate the relevance between each feature, negating the feature variables with a strong correlation and then combining with a Genetic Algorithm (GA) and a Back Propagation (BP) Neural Network to build a regression prediction model. The proposed approach has been termed SP-GA-BP. The results showed that the proposed method can predict well with R² = 0.99519 MAE = 0.18926 MSE = 0.0072213 RMSE = 0.084978, thereby indicating that the Temperature, CO₂ Pressure, and Corrosion Inhibitor efficiency can affect the corrosion rate efficaciously. Furthermore, with the introduction of external interference, the results exhibited a high level of precision. The proposed method and the obtained results may provide a good reference value for oil pipeline maintenance.