Research On Forecast Of Energy Consumption In Liaoning Province Based On PSO-SVM Method

Energy is an important material foundation for economic growth and social development,and it is also an important guarantee for national and regional economic development.In the 40 years since the reform and opening up,China's economy has achieved rapid growth,with an average annual growth rate of 9.5%,and its proportion in the global economy has increased from 1.8%to 15.2%.However,due to the accelerated pace of economic development,energy consumption has also increased,and China has become the second largest energy consumer in the world.At the same time,the coal-based energy consumption structure has caused China's rapid economic development and caused serious environmental pollution.As an old industrial base in Northeast China,Liaoning Province is rich in resources and has huge energy consumption.In recent years,the contradiction between supply and demand in Liaoning Province has been serious,and energy needs to be imported from other provinces.At the same time,Liaoning Province has a relatively simple energy consumption structure,and its energy utilization rate is far below the national average,which has severely restricted the economic development of Liaoning Province.Affected by the economic transformation in 2015,Liaoning Province's total economic ranking dropped from the second place in 1978 to the tenth in 2015.In 2015,Liaoning's nominal GDP growth rate was-2.0%.Therefore,how to accelerate the transformation of the economic development model,take the low-carbon development path and achieve sustainable development is the first consideration of Liaoning Province.Among the overall development goals proposed in the Liaoning Province's"Thirteenth Five-Year Plan" Comprehensive Implementation Plan for Energy Conservation and Emission Reduction,one of them is:by 2020,the energy consumption per unit of GDP will be reduced by 15%compared with 2015,and the increase in energy consumption will not more than 35.5 million tons of standard coal.In order to achieve balanced development of energy,economy and environment in Liaoning Province,and to achieve the energy-saving targets set out in the "Thirteenth Five-Year Plan" for energy conservation and emission reduction in Liaoning Province,it is necessary to analyze the main factors affecting energy consumption and predict the future energy consumption situation in Liaoning Province.Based on this,a scientific and reasonable energy plan and related policies are formulated.This paper studies the current situation of the increasingly prominent contradiction between energy supply and demand in Liaoning Province.Based on a qualitative analysis of energy consumption in Liaoning Province and its influencing factors,the grey correlation method is used to quantify the influencing factors of total energy consumption.Using the energy and economic and social statistics data of Liaoning Province from 1984 to 2017,the traditional Gray Model(GM)and Support Vector Machine(SVM)model were constructed.And in order to find the optimal combination of SVM parameters,the Particle Swarm Optimization(PSO)algorithm was used to optimize the SVM and predict the energy consumption of Liaoning Province in 2020.Finally,the forecast results are compared with the energy conservation and emission reduction targets of the "Thirteenth Five-Year Plan" of Liaoning Province,and policy recommendations are put forward on this basis.The research in this paper mainly draws the following conclusions:?GDP has the largest impact on changes in energy consumption,followed by level of urbanization,consumption level of residents,industrial structure,population size,energy consumption structure,energy intensity.The grey correlation method was used to quantitatively analyze the relationship between each indicator and energy consumption.It was found that the correlation between the seven indicator variables and energy consumption was significant.The grey correlation was ranked as:GDP>level of urbanization>consumption level of residents>industrial structure>population size>energy consumption structure>energy intensity;?The PSO-SVM model has the highest prediction accuracy and The PSO-SVM model is more stable.Based on the data of Liaoning Province from 1984 to 2010,the GM,SVM model and PSO-SVM model were established,and the energy consumption of Liaoning Province from 2011 to 2017 was predicted to verify the accuracy of the model.The verification results show that the GM prediction result is the worst,the average relative error is 11.49%,the prediction accuracy is unstable.And when the GM is used for trend extrapolation,we find that the growth trend of the model will become more and more obvious,resulting in the model's later predicted value will be higher than the actual value;the average relative error of the SVM prediction model is 5.95%;the average relative error of the PSO-SVM prediction model is 4.19%,and the model has the best prediction accuracy and stability;?If the economic and social indicators involved in the calculation of the model are developed in accordance with the goals set in the planning document,using the PSO-SVM model to predict energy consumption,we can conclude that the total energy consumption in Liaoning in 2020 is 216.19 million tons and the increase in energy consumption controlled in the range of 35.5 million tons.Under this forecast scenario,the energy consumption per unit of GDP in Liaoning Province will be 2.04 in 2020,which will be 26.6%lower than in 2015.Reached the "Thirteenth Five-Year" target value for energy saving and emission reduction,with a target value of 15%.The research in this article has the following innovations:?At present,most domestic scholars' research on energy consumption prediction is based on the whole country,and few scholars make predictions on energy consumption in provinces and cities.This article is aimed at the current energy consumption situation in Liaoning Province.It is a forecast study of Liaoning Province's energy consumption in 2020,and measures whether Liaoning Province's "Thirteenth Five-Year Plan" energy saving target has been achieved;?At present,more researches explore the correlation between energy consumption and one or two influencing factors.This article uses qualitative and quantitative(gray correlation analysis)methods to select GDP,level of urbanization,industrial structure,consumption level of residents,energy consumption structure,population size,energy intensity as the influencing factors affecting energy consumption.And use the above factors to build a model for prediction;?A new energy consumption prediction model is constructed.SVM has also been widely used in prediction,and the SVM model has better prediction accuracy for small sample data,but now there are not many studies on energy consumption using SVM model.And the SVM model has the problems of being sensitive to parameter values and difficult to determine the parameters.In this paper,we use PSO to iteratively find the optimal parameter combination of SVM(penalty parameter C,insensitivity coefficient p,and kernel width d),and construct an energy consumption prediction model based on PSO-SVM with high prediction accuracy.