| description abstract | Corrosion is the main reason for the failure of CO2 pipelines. The corrosion behavior and mechanism of CO2 pipelines significantly differ from those of traditional oil and gas pipelines. The existing CO2 pipeline corrosion prediction models have low prediction accuracy. Given the presented problems, first, considering the interaction of corrosion influencing factors such as impurities, the kernel principal component analysis (KPCA) is used to reduce the dimension of the factors affecting the internal corrosion behavior of CO2. After that, 27 groups of internal corrosion rate test data were used as training sets. The backpropagation (BP) neural network algorithm was used to construct a prediction model of the pipeline internal corrosion rate based on KPCA–BP. Finally, the reliability of the model constructed in this paper and the existing corrosion prediction model is verified using eight sets of test set corrosion data. The results show that the KPCA algorithm uses the cumulative contribution rate as the evaluation index to determine five principal components, effectively reducing the prediction model’s complexity. The root mean square error, average absolute percentage error, and determination coefficient of the KPCA–BP combined model for predicting the internal corrosion rate of CO2 pipeline are 0.00735, 20.18, and 0.9857, respectively, and the average relative error is 15.95%. Compared with the existing corrosion prediction models, the prediction accuracy of the KPCA–BP combined model for the internal corrosion rate of the CO2 pipeline is greatly improved. It lays a foundation for the subsequent evaluation of residual strength and residual life of the CO2 pipeline. | |
