Accurately predicting the power consumption of crude oil pipelines is conducive to controlling the energy consumption level of such pipelines and fully tapping the energy saving potential of crude oil pipeline transportation systems. Actual operation data of such pipelines have the characteristics of large fluctuation range serious noise interference, and information redundancy, which affect the accurate prediction of pipeline power consumption. To solve these problems, this paper proposes a power consumption prediction model based on a hybrid neural network. The daily operation data of crude oil pipelines are decomposed by complete ensemble empirical mode decomposition with adaptive noise. Principal component analysis is performed to reduce the dimensions of the decomposed data. The improved particle swarm optimization algorithm is applied to adjust the structural parameters of the neural network. The proposed model is applied to predict the power consumption of a crude oil pipeline and compared with some common prediction models. The results show that the decomposition algorithm can improve the prediction accuracy of the model. The hybrid neural network model has the highest prediction accuracy. The average absolute error of the test set is 5.394%, which is 39.200% lower than that before the decomposition algorithm is used.