基于PSO-LSTM网络模型的建筑碳排放峰值预测

Forecast of Peak Carbon Emissions of buildings Based on PSO-LSTM model

摘要:

对建筑领域CO2排放进行控制是我国实现“双碳”目标的关键抓手。我国建筑碳排放量由1990年的44 661.73 万吨增长至2019年的20 556.03 万吨，年均增长率达12.01%，具有巨大的碳减排潜力。本文对BP神经网络、LSTM网络和PSO-LSTM模型在碳排放预测方面进行对比选优后，通过训练好的PSO-LSTM模型在低碳、基准、高碳三种情景下，分别对建筑碳排放峰值进行了预测。结果表明，低碳、基准、高碳三种情景的建筑碳排放峰值分别为226 774.56 万吨、239 738.11 万吨和253 379.47 万吨；达峰时间分别为2029年、2032年和2033年。可见，在现有社会发展背景下，仍难在2030年前实现建筑领域的碳达峰，还需采取相应的低碳措施来推进目标的实现。

Abstract:

The control of CO2 emissions in the construction sector is essential to achieve the ‘Double Carbon’ target in China. China's carbon emissions in construction sector of operation have increased from 44 661.73 million tons in 1990 to 205 560.03 million tons in 2019, with an average annual growth rate of 12.01%. The potential of construction sector to contribute to the carbon emission reduction target can not be ignored. In this paper, after comparing and selecting BP neural network, LSTM network and PSO-LSTM model for carbon emission prediction, the trained PSO-LSTM model was used to predict the peak carbon emission of buildings under three scenarios of low carbon, baseline and high carbon, respectively. The results show that the peak building carbon emissions for the low-carbon, baseline and high-carbon scenarios are 226 774.56 million tons, 239 738.11 million tons and 253 379.47 million tons, respectively; and the peak time is 2029, 2032 and 2033, respectively. It can be seen that in the existing social development context, it is still difficult to achieve the carbon peak in the building sector by 2030, and corresponding low-carbon measures are needed to promote the achievement of the goal.