Natural Gas Consumption Prediction

The estimation on natural gas consumption is one of the most important decision-making for the layout, design, optimization and control of net pipeline's transportation as well as admeasure. This thesis mainly takes actual condition of gas consumption as object, combined with its historical data. On the basis of these, it is analyzed the factors that affect gas consumption, and is brought forward the methods of gliding average and trends joint, in order to pretreat the historical data of long and short term consumption. Meanwhile, their mathematical models are established. For long term one, it could be forecasted by the method of multivariate regression and artificial neural network. According to the result programmed by VB language, BP neural network is the best way, with a precision scope from 1.01% to 8.23%. For short term one ,it is the methods of AR model and trend percentage-Wentes combination that were applied to forecast natural gas consumption. According to the result, trend percentage-Wentes combination is the best way, with a precision scope from 0.78% to 2.37%.