基于时间序列的新能源汽车销售量预测——以比亚迪为例  

Time Series-based Forecasting of New Energy Vehicle Sales:Taking BYD as an Example

作　　者：邹瑞 刘吉华[1] 许思为 ZOU Rui;LIU Jihua;XU Siwei(Business College,Hubei University,Wuhan 430062,China)

机构地区：[1]湖北大学商学院,武汉430062

出　　处：《科技和产业》2024年第15期87-93,共7页Science Technology and Industry

摘　　要：新能源汽车的发展对于推进“双碳”目标实现起着关键作用,准确预测销量对于政策制定和企业发展有着重要意义。以比亚迪新能源汽车作为研究对象,运用其历史销量数据分别构建季节性自回归差分移动平均(SARIMA)和长短期记忆(LSTM)网络预测销量。为提升模型预测效果,集成单一模型得到ARIMA-LSTM(自回归差分移动平均-长短期记忆)组合模型,将销量数据分解为线性和非线性两部分,使用ARIMA模型预测销量数据中的趋势,模型的残差及其余非线性部分的数据使用LSTM模型预测,最终将模型的预测结果合并。将组合模型应用于国内新能源汽车销量预测,预测精度为90.96%,效果较单一模型有显著提升。The development of new energy vehicles plays a critical role in achieving the“dual carbon”goals.Accurate sales forecasting is of great importance for policy-making and corporate growth.BYD’s new energy vehicle is used as the research object,and seasonal autoregressive integrated moving average(SARIMA)and long short term memory(LSTM)models are constructed based on their historical sales data to forecast future sales.To improve model performance,an ARIMA-LSTM(autoregressive integrated moving average and long short term memory)hybrid model is created.Sales data is decomposed into linear and nonlinear parts.The ARIMA model is used to forecast trends,while the LSTM model is applied to predict residuals and other nonlinear data.The final results from both models are combined.This hybrid model is applied to forecast domestic new energy vehicle sales,and its accuracy is 90.96%,showing significant improvement over single-model predictions.

关 键 词：汽车销量预测 季节性自回归差分移动平均(SARIMA) 神经网络 新能源汽车 

分 类 号：F416[经济管理—产业经济]