%0 Journal Article
%T 基于ARIMA的多元线性回归模型的全球气候变暖分析<br>Global Warming Analysis Based on ARIMA Multiple Linear Regression Model
%A 周雨蕙
%A 唐旗英
%J Modeling and Simulation
%P 1257-1271
%@ 2324-870X
%D 2023
%I Hans Publishing
%R 10.12677/MOS.2023.122118
%X 全球气候变暖已成为国际社会广泛关注的问题，本文利用ARIMA和多元线性回归模型来分析全球的温度变化和影响气温的因素。首先我们对1992年到2022年全球平均气温的12个月份数据，使用SPSS软件进行可视化与预处理，对异常值的波动进行总结并建立数学模型。然后我们开发一种机器学习方法对火山爆发、森林火灾、新冠肺炎和太阳活动异常四种自然灾害的影响进行分析，给出了长期线性趋势，得出自然灾害对气候变暖的影响。接着对于影响全球气温变化的原因，我们从降水量、海洋表面温度、全球平均温度、二氧化碳浓度、地球吸放热等因素进行考虑，利用逐步回归法通过SPSS进行相关性检验求解多元线性微分方程，得出地球放热对温度影响最大。最后我们对模型进行了优化和推广，将ARIMA模型与多元线性回归预测模型进行结合，利用极寒天气情况对模型进行推广，检验模型的灵敏度，最终的拟合效果较好，应用场景较广。<br />
Global warming has become a concern of the international community. In this paper, ARIMA and multiple linear regression models are used to analyze the global temperature change and the fac-tors that affect the temperature. Firstly, the 12-month data of global mean temperature from 1992 to 2022 were visualized and preprocessed by SPSS software, and the fluctuations of outliers were summarized and mathematical models were established. Then, we developed a machine learning method to analyze the impact of four natural disasters, volcanic eruption, forest fire, COVID-19 and abnormal solar activity, and gave a long-term linear trend to get the impact of natural disasters on climate warming. Then, for the reasons affecting the change of global temperature, we considered precipitation, ocean surface temperature, global average temperature, carbon dioxide concentra-tion, earth heat absorption and release and other factors. We used the step regression method to solve the multivariate linear differential equation by SPSS correlation test, and concluded that the earth heat release had the greatest influence on temperature. Finally, we optimized and extended the model, combined the ARIMA model with the multiple linear regression prediction model, pro-moted the model with the extreme cold weather, and tested the sensitivity of the model. The final fitting effect was good and the application scenarios were wide.
%K 全球变暖，机器学习方法，ARIMA，多元线性回归模型；Global Warming
%K Machine Learning Methods
%K ARIMA
%K Multiple Linear Regression Model
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=62793