Abstract: With the deepening development of oilfields and the application of diversified oil extraction methods, the corrosion environment of the surface gathering and transportation system becomes increasingly complex, causing frequent leakage accidents due to corrosion damage to crude oil gathering and transportation pipelines. In this study, by establishing numerical simulation calculations of corrosion failure in water-injection pipelines within the crude oil gathering and transportation system, considering the synergistic influence mechanism between erosion-corrosion and electrochemical corrosion, using 11 pipeline characteristic parameters as corrosion failure features, and erosion-corrosion rate, electrochemical corrosion rate, and comprehensive corrosion rate as prediction indicators, a dataset integrating corrosion mechanisms and characteristic data for crude oil gathering and transportation pipelines was constructed. Using the BP neural network optimized by Kohonen and MEA algorithms, prediction models driven by mechanism and data fusion for erosion-corrosion rate, electrochemical corrosion rate, and comprehensive corrosion rate of crude oil gathering and transportation pipelines were established and compared against various prediction models. The result shows that the optimized Kohonen clustering algorithm, taking pipeline specifications and laying methods as potential dominant features, divides the corrosion failure into four levels according to different pipeline characteristics. The Kohonen-MEA-BP model driven by mechanism and data performs best in predicting the erosion-corrosion rate, electrochemical corrosion rate, and comprehensive corrosion rate of crude oil gathering and transportation pipelines, with quantitative evaluation indicators MSE of 0.001 2, 4.8×10^-6, and 0.000 3, respectively. Besides, the model reveals the variation patterns of erosion-corrosion rate and comprehensive corrosion rate corresponding to different corrosion failure levels, indicating that the corrosion failure coupling mechanism of water-injected pipelines is dominated by erosion-corrosion. This study provides a reference for the integrity management of crude oil gathering and transportation pipelines under the development of digital construction.