Forecasting New Energy Vehicle Sales Using Improved BP Neural Network and SARIMA Models
Qi‐Fu Chen, Zijie Hong, Xi Chen, Xiaqiang You · 2024
To improve the prediction accuracy of standard BP neural networks for both non-random and random fluctuation sequences in new energy vehicle sales data, an enhanced prediction model is proposed. This model initially utilizes the SARIMA model to forecast the non-random fluctuation sequence in the sales data. Then, a BP neural network model is applied to predict the random fluctuation sequence. The final sales prediction for new energy vehicles is obtained by combining these two sets of predictions. Simulation experiments were conducted to validate this approach, and the results indicate that the improved BP neural network method can effectively forecast the sales trend of new energy vehicles, achieving an average prediction accuracy of 90.82%. This makes it suitable for practical monthly sales forecasting of new energy vehicles.