Research On Gas Load Forecasting Technology Based On Combined Model

With the rapid development of the gas industry,the prediction of gas load has become an important task of the gas system management department.It is of great significance to improve the economic and social benefits of Gas Co enterprises and to maintain the safe and stable operation of the gas system by grasping the characteristics and changing rules of the gas load and predicting the gas load accurately and reasonably.Therefore,in order to meet the growing demand of urban gas pipeline network planning,operation and maintenance management and gas storage peak adjustment,it is necessary to explore a more accurate and more applicable method to predict the gas consumption reasonably so as to alleviate the contradiction between supply and demand,and effectively solve the problem of gas supply in the development of China.For this reason,the following research work has been carried out in this paper.First,the gas load characteristics of Shaanxi province and Nanye county are analyzed in detail.According to the characteristics of load change,the influence factors of medium and long term and short-term gas load are excavated.Secondly,considering the principles and advantages and disadvantages of principal component analysis(PCA),grey model(GM)and BP neural network model(BPNN),a new PCA-GM-BPNN combined prediction model for gas load is proposed.Thirdly,according to the grey theory and its optimization model,BP neural network and its optimization model and the new PCA-GM-BPNN combination forecasting model,the gas annual load of Shaanxi Province from 2004 to 2011,the daily load data of Nanle county from September 1st to October 20 th and the hourly load data of Nanle county from October 1st 0:00 to October 15 th 24:00 are established to predict the annual load data of Shaanxi province from 2012 to 2015,the daily load data of Nanle county from October 21 st to October 30 th and the hourly load data of Nanle county from October 16 th 0:00 to October 18 th 24:00 by MATLAB software.The forecast results show that the annual load forecast MAPE value of the PCA-GM-BPNN combined forecasting model is 8.21%,the daily load forecasting MAPE value is 5.44%,the hour load forecasting MAPE value is 6.98%,which is less than the other four load forecasting models,which fully shows the superiority of the PCA-GM-BPNN combination model in the year,day and hour load forecasting.Finally,the GLP gas load prediction software is developed,and the Java program based on grey theory algorithm and BP neural network algorithm is compiled.