Study On Influencial Factors Of Carbon Emissions Based On LSSVM Model Optimized By Modified Firefly Algorithm

Energy consumption is always an important driving force for the human economic development and social progress.For China,the fossil fuels like coal and crude oil has dominated the production and life of the people of the country.However,under the background of the more serious global climate change and the rise of the low-carbon sustainable economy development,the issue about the carbon dioxide emissions which are caused by the using process of energy consumption can not be ignored.Thus,as the first developing country,China need to ensure the security of energy supply,reduce the carbon emissions,cope with the climate change and undertake the obligations and responsibilities of international environmental protection and simultaneously China should guarantee the sound and fast development of the domestic economy.This has become a major problem to be solved urgently.Hebei Province,as a major province in China,its environmental problems are extremely severe in recent years,especially the haze pollution.Taking into account that the carbon emissions are a major factor,it is very significant to study the influencial factors of the carbon emissions in Hebei province.In this paper,the concepts and theories related to the carbon emissions and the current situation of China's carbon emissions are introduced first as well as the concrete domestic and foreign research on carbon emissions.Secondly,the least squares support vector machine(LSSVM)which is a algorithm suitable for small sample data is presented.Considering the penalty factor and kernel function width are determined empirically,a heuristic intelligent algorithm--the modified firefly algorithm(MFA)is used to search the optimal value in order to increase the accuracy of the algorithm.In the section of case study,this paper selects the relevant data about carbon emissions and influencial factors of Hebei province in 1990-2014 as the research object.The SPSS statistical software is used to compute the significant correlation among the pre-selected factors.Then the proposed algorithm is utilized for a pertinent modeling according to the characteristics of the sample data.To measure the specific impact of factors on carbon emissions quantitatively and better,the stochastic impacts by regression on population,affluence and technology(STIRPAT)and the logarithmic mean divisia index(LMDI)model are also applied to analyze the relationship between carbon emissions and impact factors.Based on the results,the following conclusions could be obtained:(1)The algorithm presented in this paper verifies the causal relationship between carbon emissions in Hebei province and the thirteen influencing factors identified by SPSS screening.And the new method is validated superior to three other algorithms,which proves the modification to the standard FA and optimization to the LSSVM are effective.(2)The STIRPAT model shows that the the effects of 13 factors to carbon emissions are positive correlation.Among them,the driving index of the final consumption is the largest while that of the traffic vehicle ownership is the smallest.(3)The LMDI method is used to decompose six factors,including the carbon emission coefficient,energy intensity,energy consumption structure,industrial structure,economic activity scale and population size.And the result is the economic activity is the biggest factor to promote the carbon emissions in Hebei province among all factors.Finally,this paper puts forward some specific suggestions from the aspects of energy structure,energy efficiency,population policy,transportation and low-carbon concept,which is in favor for the government to formulate emission reduction policies in view of the theoretical support and benefit to control the production of carbon emissions from the source effectively.