为应对极端气候,贯彻新发展理念,助力碳达峰碳中和,本论文基于多元线性回归模型和LSTM神经网络模型建立碳排放量的预测模型,为实现双碳目标提供路径规划。本文所做的工作能够概括如下:首先,筛选出经济、人口、能源排放量和碳排放量的主...为应对极端气候,贯彻新发展理念,助力碳达峰碳中和,本论文基于多元线性回归模型和LSTM神经网络模型建立碳排放量的预测模型,为实现双碳目标提供路径规划。本文所做的工作能够概括如下:首先,筛选出经济、人口、能源排放量和碳排放量的主要指标并建立指标体系;其次,采用斯皮尔曼相关系数分析得出各因素对碳排放量的贡献程度,利用多元线性回归建立区域碳排放量与经济、人口、能源消费量各指标的关联模型。再次,采用LSTM神经网络模型预测2021~2060年的该区域碳排放量及各部门碳排放量。最后,选择同样的方法建立基于LSTM的与各部门能源品种相关的碳排放量预测模型。In order to cope with extreme climate, promote high-quality development, and contribute to achieving carbon peak and carbon neutral, this thesis establishes a prediction model for carbon emissions based on multiple linear regression model and LSTM neural network model to provide path planning for achieving the dual carbon target. The work done in this paper can be summarized as follows: firstly, the main indicators of economy, population, energy emission and carbon emission are screened out and the indicator system is established;secondly, the Spearman correlation coefficient analysis is used to derive the degree of contribution of each factor to the carbon emission, and the multiple linear regression is used to establish the correlation model between the regional carbon emission and the indicators of economy, population and energy consumption. Again, the LSTM neural network model was used to predict the carbon emissions of the region and the carbon emissions of each sector from 2021 to 2060. Finally, the same method was chosen to establish LSTM-based prediction models of carbon emissions associated with energy varieties in each sector.展开更多
文摘为应对极端气候,贯彻新发展理念,助力碳达峰碳中和,本论文基于多元线性回归模型和LSTM神经网络模型建立碳排放量的预测模型,为实现双碳目标提供路径规划。本文所做的工作能够概括如下:首先,筛选出经济、人口、能源排放量和碳排放量的主要指标并建立指标体系;其次,采用斯皮尔曼相关系数分析得出各因素对碳排放量的贡献程度,利用多元线性回归建立区域碳排放量与经济、人口、能源消费量各指标的关联模型。再次,采用LSTM神经网络模型预测2021~2060年的该区域碳排放量及各部门碳排放量。最后,选择同样的方法建立基于LSTM的与各部门能源品种相关的碳排放量预测模型。In order to cope with extreme climate, promote high-quality development, and contribute to achieving carbon peak and carbon neutral, this thesis establishes a prediction model for carbon emissions based on multiple linear regression model and LSTM neural network model to provide path planning for achieving the dual carbon target. The work done in this paper can be summarized as follows: firstly, the main indicators of economy, population, energy emission and carbon emission are screened out and the indicator system is established;secondly, the Spearman correlation coefficient analysis is used to derive the degree of contribution of each factor to the carbon emission, and the multiple linear regression is used to establish the correlation model between the regional carbon emission and the indicators of economy, population and energy consumption. Again, the LSTM neural network model was used to predict the carbon emissions of the region and the carbon emissions of each sector from 2021 to 2060. Finally, the same method was chosen to establish LSTM-based prediction models of carbon emissions associated with energy varieties in each sector.
