Deep Learning Combinatorial Models for Intelligent Supply Chain Demand Forecasting
Xiaoya Ma, Mengxiu Li, Tong Jin, Xiaying Feng
Abstract
Low-carbon and environmentally friendly living boosted the market demand for new energy vehicles and promoted the development of the new energy vehicle industry. Accurate demand forecasting can provide an important decision-making basis for new energy vehicle enterprises, which is beneficial to the development of new energy vehicles. From the perspective of an intelligent supply chain, this study explored the demand forecasting of new energy vehicles, and proposed an innovative SARIMA-LSTM-BP combination model for prediction modeling. The data showed that the RMSE, MSE, and MAE values of the SARIMA-LSTM-BP combination model were 2.757, 7.603, and, 1.912, respectively, all of which are lower values than those of the single models. This study therefore, indicated that, compared with traditional econometric forecasting models and deep learning forecasting models, such as the random forest, support vector regression (SVR), long short-term memory (LSTM), and back propagation neural network (BP) models, the SARIMA-LSTM-BP combination model performed outstandingly with higher accuracy and better forecasting performance.