数学建模方法对碳排放、碳达峰问题的研究  

Research on Carbon Emissions and Carbon Peaking by Mathematical Modeling Methods

作　　者：李伟 梁旭 王宇峰 Wei Li;Xu Liang;Yufeng Wang(School of Mathematics and Statistics,Xidian University,Xi’an Shaanxi)

机构地区：[1]西安电子科技大学数学与统计学院,陕西西安

出　　处：《建模与仿真》2024年第4期5001-5012,共12页Modeling and Simulation

摘　　要：以河北省碳排放为研究目标,本文基于传统KAYA恒等式和STIRPAT模型,提出了一个改进的STIRPAT模型,并用于预测河北省碳排放以及碳达峰情况。首先使用ARIMA模型、灰色预测方法、长短期记忆(LSTM)网络预测模型分别对改进STIRPAT模型中的驱动指标进行了预测和比较,得到了各指标的最优预测方法。之后,结合时序型预测方法,分别采用BP神经网络回归、Lasso回归和岭回归方法分别对STIRPAT模型种的碳排放进行了预测。结果表明,随着第三产业占比的持续扩大和清洁能源占比的增加,河北省的碳排放总量将呈现下降趋势。其中Lasso回归方法具有最小均方误差和最优拟合优度,这种回归方法预测出河北省的碳排放总量在2035年将达到456.8百万吨,并在2023年左右达到碳排放峰值。河北省的碳排放预测对同类省份的碳排放情况给出了参考,这对于中国全国实现2035年碳达峰目标具有重要意义。Taking carbon emissions in Hebei Province as the research objective,this paper proposes an im-proved STIRPAT model based on the traditional KAYA identity and STIRPAT model,and uses it to predict carbon emissions and carbon peaking for Hebei Province.Firstly,the ARIMA model,Grey prediction method,and Long-Short Term Memory(LSTM)network were used to predict and compare the indicators in the improved STIRPAT model,and the optimal prediction method for each indicator was obtained.Afterwards,combined with time-series prediction methods,BP neural network regression,Lasso regression and ridge regression methods were used to predict carbon emissions in the STIRPAT model,respectively.The results indicate that with the continuous ex-pansion of the proportion of the tertiary industry and the increase of the proportion of clean energy,the total carbon emissions in Hebei Province will show a downward trend.The Lasso re-gression method has the minimum mean-square error and optimal fitness.This regression me-thod predicts that the total carbon emissions in Hebei Province will reach 456.8 million tons by 2035 and reach carbon peaking around 2023.The carbon emission forecast of Hebei Province provides a hint for the carbon emission of similar provinces,which is of great significance for Chi-na to achieve the carbon peaking target by 2035.

关 键 词：碳排放预测 改进STIRPAT模型 时序型预测方法 回归型预测方法 

分 类 号：F42[经济管理—产业经济]