Research On The Macro Model Of City Gas High Pressure And Medium Pressure Pipeline Network

With the high-speed development of our country in recent years, natural gasdemand is increasing and the use of natural gas is gradually into the variousindustries. As gas downstream suppliers, city gas companies transport anddistribution of natural gas directly to end users. It is an important task of gascompanies’ operating to ensure the end user gas supply safety and economy. How toanalysis the hydraulic conditions of pipe network quickly and predict the nextmoment pressure of control points is the key of gas supply. Macroscopic model canavoid the study of pipe network structure parameters and status parameters. It onlystudies the relationship of the operating parameters of the city gate stations, storageand distribution stations and pipe network control points. It is easy, fast and suitablefor the real-time scheduling of city gas.Before performing modeling, station flow is needed to predict. It is the mostimportant inputs variable of the model. Both complete historical data and parts ofhistorical data in the prediction are considered in the paper. The exponentialsmoothing method is used to predict the station flow with complete historical datapipe network system. And typical load curve is used to solve the prediction withonly parts of historical data. In the paper, the daily load forecasting by exponentialsmoothing method is validated with data of A city. And a new method to determinetypical gas load curve with correlation analysis is proposed.This article based on the existing pipeline network regression model, the linearregression model is proposed for the gas pipeline network having a plurality of thecity gate stations. In order to verify the correctness of the model, a case is analyzed.Because of some outliers may lead to large errors, so the multiple linearperipheral regression model is proposed in this article. The multiple linearperipheral regression model is applied to the city gas network pressure warning andcapacity calculations. For the gas peak prediction accuracy is not high, the gas peakquadratic regression model is establish separately and it dramatically improves theprediction accuracy.The regression model obtained a good prediction accuracy throughsegmentation modeling However, there are still limitations in the regression model.The limitations include the number of multi-regression equation, poor faulttolerance, narrow range of applications of the equation. In response to theselimitations, this paper proposes a BP neural network modeling of macro ideas. Artificial neural network is regarded as one of the most promising areas in artificialintelligence in the20th century world. BP artificial neural network is used in thisarticle, the gate station flows and macroeconomic variables as inputs, and the BPnetwork architecture is used in the gas pipeline. By the compiler, the H city gaspipeline network is conducted the MATLAB simulation.And the predictionaccuracy is high. Both gas peak hours and gas valley hours the model can get a goodprediction accuracy. Compared with the regression model, BP artificial networkmodel has fewer macro models, broader applicable scope and higher accuracy.In the macro model, the flow of city gate station is input to the model and thepressure of control point is the output. Only city gate station flow projections areaccurate, the macro model can output the correct control point pressure. In the paper,the daily load forecasting by exponential smoothing method is validated with dataof A city. And a new method to determine typical gas load curve with correlationanalysis is proposed.