Abstract: Energy consumption of oil pipeline is affected by many factors. To predict the energy consumption accurately, the BP neural network, which is capable of self-organizing, self-adaptation, and approximating any nonlinear continuous mapping, is adopted to build a prediction model. In order to make the model more generalized, the error control formula is added in the computing process with traditional BP neural network. Finally, a prediction model of crude oil pipeline is obtained based on improved BP neural network. This model is applied to the samples of energy consumption taken from a pipeline, which are pre-processed in order to improve the convergence speed and accuracy. The results show that the simulation error of the model is within 2.77%, and the analog value can truly reflect the true value. Then, the model is promoted to a natural gas pipeline, indicating that the model can accurately predict the energy consumption of natural gas pipeline, with error not more than 4.06%. Therefore, this model, as a new method, is applicable to predicting the energy consumption of oil and gas pipelines.