Forecasting Of Natural Gas Load Based On Grey-Neural Theory

The commercial operation of the west-east natural gas pipeline project and Sichuan to EastGas Pipeline Project signifies that the arrangement of the national gas pipeline network starts tochange greatly. The respective city’s single network system has been gradually changing to thelarge-scale national network system. It’s a urgent problem for realizing effectively run,optimaloperation and scientific administration of city gas pipeline network system.to grasp the regularityof gas load and get accurately and reasonably predicted value.First of all, this paper researches the regularity and influential factors of gas load fromhistorical data which included economy, population, the usage of liquefied petroleum gas,weather, holidays and so on. GM (1,1) model, dynamic reform carry model, BP artificial neuralnetwork model were been built by MATLAB tool. In order to improve the generalized characterand the predicted precision of neural network, correlation analysis, partial correlation analysis,principal component analysis were used to qualitative study the impact factor of gas load in orderto find the main factors of the city gas load. Optimizing influential factors,unit number ofhidden layer, transferring function and learning function were used to build the BP neuralnetwork optimization model; What’s more, the optimal model was selected via grey correlationanalysis and other statistical methods. Grey-neural network gas load forecasting model wasestablished with the optimal model by variance-covariance optimization combination forecastingmethod.Prediction models were built with indicators in the range of1997and2011of shaanxiprovince and in the range of2013October of xi ’an. The Predictive value of gray-neural networkmodel show that relative error of the middle-long term gas load is between0.41%and3.9%, theshort-term gas load is between0.27%and1.72%, compared with other models, the gray-neuralnetwork model has more higher precision and more reasonable.