Oil spill area prediction model of submarine pipeline based on BP neural network and convolutional neural network

In recent years, there has been a growing demand for oil resources. The increasing development of marine oil transport and exploration also brings the problem of marine oil spill. Oil spill prediction models can be used to predict the spreading behaviour of oil spills, which is an important tool for risk assessment during oil spill accidents. Therefore, by calculating the oil spill area under different water depth (static water), leakage aperture, sea surface wind speed and 0# diesel pipeline flow rate within 0.5–7 s, a total of 686 sets of oil spill area data were generated. Four algorithms, namely BP neural network, genetic algorithm-optimized BP neural network, particle swarm optimization BP neural network and convolutional neural network, were employed to predict the oil spill area, and a prediction model for the oil spill area of submarine pipelines was established. The influencing factors such as water depth, leakage aperture, sea surface wind speed and pipeline flow velocity were used as model inputs, and the output was the prediction result. By comparing the training and validation results of BP,PSO-BP,GA-BP and CNN, it was found that the PSO-improved BP neural network prediction model had a higher prediction accuracy for the oil spill area of submarine pipelines. Compared with the ordinary BP neural network, the RMSE of the test set was reduced by 54 %, and the MAE was reduced by 50.22 %, which basically met the practical application standards. It is determined that the PSO-BP neural network embodies better convergence and accuracy in oil spill area prediction of oil film, so the PSO-BP neural network model can be used as the prediction of oil spill area, which can provide the theoretical basis and technical support for the reliable prediction of oil spill accidents.