Abstract: In order to improve the prediction accuracy of internal corrosion rate of submarine oil and gas pipelines, the corrosion data of a mixed transport pipeline section in the South China Sea were selected as an example to establish the internal corrosion rate prediction model based on core principal component analysis（KPCA）, seagull optimization algorithm（SOA） and kernel extreme learning machine（KELM）. Firstly, the KPCA was used to reduce the dimension of the factors affecting the corrosion in the pipeline, and the input variables were determined. Then, KELM was used to model and predict the internal corrosion rate, and SOA was used to optimize the kernel parameters and regularization coefficient in KELM model. The results show that the mean absolute percentage errorof KPCA-SOA-KELM prediction model is only 1.831%, and the root mean square erroris 0.05. Compared with other models, the prediction results of KPCA-SOA-KELM prediction model are more accurate.