To solve the problem of pipeline leakage location determination, the leak situation of a trans-oceanic pipeline was modeled by EPANET software, and the BP neural network model and K-CV improved SVR model were used to predict the leak location. Firstly, the number of hidden layers and learning function of BP neural network were optimized and selected, and K-CV method was used to search for the optimal combination of the penalty coefficient c and the kernel function parameter g of SVR. The EPANET software modeling data was used to form a training set, and the test set was randomly selected from original data for prediction. Meanwhile, the relative mean square error and correlation coefficient were used to evaluate the prediction results. The experimental results show that K-CV method can effectively improve the accuracy of SVR model. Compared with the traditional steady-state hydraulic equation, the neural network model has a wider range of applications and higher accuracy in the prediction of leak location of trans-oceanic pipelines.