基于多源数据融合与Prophet-XGBOOST混合模型的新能源汽车渗透率预测

Prediction of New Energy Vehicle Penetration Rate Based on Multi-Source Data Fusion and Prophet-XGBOOST Hybrid Model

摘要: 在能源安全问题日益凸显与碳达峰、碳中和目标的催化下，新能源汽车成为缓解能源压力、推动绿色转型的重要力量。因此，准确预测新能源汽车渗透率对产业可持续发展具有重要意义。本文运用多源数据融合技术，选取2023年1月到2025年12月新能源汽车月度销量数据、充电桩数据、GDP、能源价格及双积分政策文本作为基础数据，构建Prophet-XGBoost模型进行预测。利用Prophet模型提取渗透率趋势项和季节项作为时序特征，通过XGBoost模型融合时序特征和多源外部因素进行训练与预测。对比分析表明，混合模型在测试集上的MAE和RMSE较单一Prophet模型分别降低0.12%和0.15%，为未来新能源汽车渗透率的预测提供了一个精度较高的有效模型。结果显示2026年新能源汽车渗透率有望在年底达到54.20%。

Abstract: Under the catalysis of the increasing prominence of energy security issues and the goal of carbon peaking and carbon neutrality, new energy vehicles have become an important force to alleviate energy pressure and promote green transformation. Therefore, accurately predicting the penetration rate of new energy vehicles is of great significance to the sustainable development of the industry. This paper uses multi-source data fusion technology to select the monthly sales data, charging pile data, GDP, energy price and double-point policy text of new energy vehicles from January 2023 to December 2025 as the basic data, and construct the Prophet-XGBoost model for prediction. Use the Prophet model to extract the penetration rate trend items and seasonal items as timing characteristics, and use the XGBoost model to integrate timing characteristics and multi-source external factors for training and prediction. Comparative analysis shows that the MAE and RMSE on the test set of the hybrid model are 0.12% and 0.15% lower than the single Prophet model, respectively, providing an effective model with high accuracy for predicting the penetration rate of new energy vehicles in the future. The results show that the penetration rate of new energy vehicles in 2026 is expected to reach 54.20% by the end of the year.