ORIGINAL RESEARCH
Carbon Emissions Scenario Prediction of the Thermal Power Industry in the Beijing-Tianjin-Hebei Region Based on a Back Propagation Neural Network Optimized by an Improved Particle Swarm Optimization Algorithm
Jianguo Zhou, Shijuan Du, Jianfeng Shi, Fengtao Guang
 
More details
Hide details
Department of Economics and Management, North China Electric Power University,
689 Huadian Road, Baoding 071003, China
Online publish date: 2017-07-05
Publish date: 2017-07-25
Submission date: 2016-12-10
Final revision date: 2017-01-23
Acceptance date: 2017-02-08
 
Pol. J. Environ. Stud. 2017;26(4):1895–1904
KEYWORDS
TOPICS
ABSTRACT
Rapid economic growth in the Beijing-Tianjin-Hebei region has been accompanied by a dramatic increase in carbon emissions. Therefore, a precise study of forecasting carbon emissions is important as regards curbing them. To identify the influence factors of carbon emissions and effectively predict carbon emissions under the three different GDP growth rate scenarios in the Beijing-Tianjin-Hebei thermal power industry, we employed a combination of the improved particle swarm optimization-back propagation algorithm (IPSO-BP) with scenario prediction. The results are as follows:
1) The influencing degree of carbon emissions factors from strong to weak are the installed capacity of thermal power, thermal power generation, urbanization rate, GDP, and utilization ratio of units (with grey correlation degrees of 0.9262, 0.9247, 0.8683, 0.8082, and 0.7704, respectively).
2) Compared with the BP neural network, it is testified that using the IPSO-BP neural network model with an annual average relative error of 2.53%, while the prediction precision of BP neural network is 5.07%. Besides, the number of iterations to achieve the optimal solution is approximately reduced by 33%.
3) GDP is the contributor to the increment of carbon emissions of the power industry, whereby GDP growth rate can be reduced appropriately to curb carbon emissions, avoiding excessive pursuit of economic growth.
eISSN:2083-5906
ISSN:1230-1485