Deep learning power grid project cost prediction under low-carbon background

Abstract

Abstract Under the background of low-carbon power system developing towards cleaner, more efficient and more economical direction, accurate prediction of power grid project cost is very important for the construction of power grid project. In order to improve the accuracy of power grid project cost prediction, a power grid project cost prediction method based on deep learning is proposed. First, for the complex and nonlinear features of the line engineering cost data, the XGBoost model is used to analyze the line engineering cost data. Feature selection and data dimension reduction are carried out through the importance ordering of the features analyzed. Then, the engineering cost integration model is established based on the XGBoost model and Stacking. The cost analysis of power grid transmission project shows that the integrated model has better generalization performance and higher prediction accuracy, and can accurately predict the line cost level.