Analysis Of Gases Use And Load Forecasting Research Of Natural Gas In-Ji’nan West City

With the development of urban modernization,natural gas has been widely used in residential life,industry and commerce,urban heating,transportation,distributed energy and other fields,It have been an indispensable and important energy for urban development.The load of urban natural gas is affected by many factors,and it is helpful to do well the dispatch planning of natural gas by grasping the law of gas consumption of all kinds of users and understanding the load structure of the company.Accurate load forecasting plays an important role in the safe operation of pipeline network and satisfying the demand of gas supply and the optimization of pipeline network.Based on the gas data of natural gas users of Shandong Jihua Gas Co.Ltd.,the paper studies and analyzes the gas consumption laws of different types of users,such as industrial,commercial,civil,heating and filling stations in Jinan.For different types of users,the gas load on the coefficient of non-uniform calculation of the coefficient range of 1.14 ~ 2.02;By temperature and non-uniform coefficient of correlation analysis,it is concluded that climate conditions is one of the important factors that influence of Ji’nan city gas load,for using the support vector machines(SVM)method for load forecasting ready based data.Based on the daily load forecasting research of natural gas in Ji’nan city,the daily average temperature,the maximum temperature,the minimum temperature,the date type,the weather change and the economic change are taken into account.The influence factors of daily load are analyzed based on the modeling prediction.In the process of modeling,the 3σ rule is used to filter the anomaly data.The influence factors such as daily average temperature,maximum temperature,minimum temperature,data type and weather condition are normalized.Using support vector machine(SVM)method,Radial basis function(RBF)kernel function is used to model,By using chinese modeling for support vector machines(CMSVM)software platform,using the grid search algorithm and cross-validation method to find the optimal parameters to get the optimal model.Finally,the natural gas daily load data of the first half of 2015 are forecasted and compared with the daily natural gas daily load data,and the relative error is within 5%,respectively.The results of the study have reference value to the company’s distribution scheduling.