The Research On The Methods Of Gas Short-term Forecasting

As a green energy resource, natural gas is the development direction of the gasin the city. Increasing the natural gas proportion in energy consumption structure isnot only good for energy conservation and emission reduction, but also can maintainthe sustainable development of economy and the society. Shanghai, one of the firstcities to use natural gas, which is greatly promoted now. especially, with thesuccessive completion of the West-East gas pipeline project has really promoted therapid development of the natural gas in Shanghai. In order to realize the efficientoperation of gas supply system, optimized dispatching and scientific management,gas load predication, the decision base, is obviously very important.Gas loadforecasting include: long-term、middle-term、short-term and very short-term loadforecasting. This dissertation emphasizes on short-term forecasting. Through theanalysis of load law of Shanghai gas system: this paper gives undivided attention tothe precise predication of gas load. It did an intensive study of different popularintelligent forecasting techniques, such as data miningtDMj^Artificial NeuralNetworktANN), Support vector machinetSVM), traditional statistics AutoregressiveIntegrated Moving Average Model and so on. In-depth analysis and research on therules and characteristics of city gas load date of the part of Shanghai, putting forwardthe idea of combination forecasting model.Through the study of city gas load timeseries we had discovered that the series of city gas load has three characteristics:trend, periodic and stochastic. According to the three characteristics, in the paper,the authors propose the idea of a short-term forecast for the city gas load thatdecompose-combination forecasting model. And three combination forecastingmethods have been proposed to verify decomposition-combined forecasting model.Before modeling gas load sequence, firstly, the method of data mining for outliermining and revision to the historical load sequence so that it can be better reflect theregularity of the gas load. Then using three methods(formula decomposition, waveletsub-band technology law and Eviews decomposition) to decompose that it has beenoutliers processing sequence. Finally, using ARIMA and BP neural network combined with sequence characteristic to model and forecast the sequence which has beendecomposed. In order to verify the validity of decomposition-combination of model,Firstly, the results that ARIMA method, BP neural network method and combinationof model based on information entropy model and forecast the gas load of city.According to the error, we found that decomposition-combination has a higheraccuracy. To further illustrate the effectiveness of decomposition-combination ofmodel, in this paper, we forecasted the short-term load on multiple dates within thesegment. Including working days, holidays and the sudden changed in weathergeneralization and specialization daily load forecast; finally, in order to illustrate theadaptability of decomposition-combination of model, in this paper, we also used thegas load which is from another gas company to modeling and forecasting. Accordingto the analysis of the error between the predicted and actual values, theexperimental results further validate the adaptability and accuracy of the modelwhich is proposed decomposition-combination in this paper.