Construction And Application Of Hybrid Time Series Prediction Models

Energy is an important material guarantee for the development of the national economy and the improvement of people's quality of life.With the acceleration of the modernization process and the increasing demand for energy,the optimization and transformation of the energy structure has become a top priority for national development.As a traditional industrial province,Shandong Province ranks first in the country in terms of total energy consumption and total coal consumption,which has led to significant energy structure problems.Shandong Province is imperative to promote the energy revolution and achieve economic transformation.Forecasting the future energy consumption of Shandong Province will help to grasp the progress of energy development in Shandong Province,help Shandong Province achieve energy structure transformation and upgrading,and then play an important role in promoting China's sustainable development and green economy.The grey prediction method is an original developed pedigree forecasting method based on the Chinese original grey theory.After combing nearly one hundred documents,we found that there is still room for improvement in the development of grey prediction models.Based on the existing research,this study:(1)Integrating the metabolic principle and nonlinear principle into the modeling strategy of the existing grey model,thus constructing the metabolic grey model(MGM)and the metabolic nonlinear grey model(MNGM);(2)Using the idea of "error correction + quadratic modeling",the international advanced ARIMA model and BP model are combined with two improved grey models,thus constructing the MNGM-ARIMA model and the MNGM-BP model.In order to measure the accuracy of the model,this study applied four new forecasting models(MGM,MNGM,MNGM-ARIMA,MNGM-BP)to the forecasting of future energy consumption in Shandong Province based on the energy consumption data period 1995-2016.The results show that the average relative errors of the four models are 6.334%,5.802%,3.477% and 3.582%,respectively.On the one hand,this results mean that four improved models' forecasting accuracy has been improved compared with the traditional grey prediction model.On the other hand,the combined model(MNGMARIMA,MNGM-BP)has higher accuracy than the single model(MGM,MNGM),which proves that the "error correction + quadratic modeling" modeling strategy is beneficial to the model.Increased accuracy.Based on four sets of prediction results,this study uses the combination weighting method of IOWGA operator theory to integrate the prediction results of various mixed models.According to the principle that the higher the accuracy is,the greater the weight coefficient is,the only predicted value with the most accuracy is finally obtained.The prediction results show that the energy demand of Shandong province in 2017-2025 will grow at an average annual rate of 7%.This result enlightens energy policy makers in Shandong province to focus on improving energy policy system,promoting coal production reform,and establishing a green and pluralistic energy supply model.