Study On Modeling Of Power System Short Term Load Forecasting Considering Multi Factor Meteorology

Short term load forecasting is focus on the next several hours,one day or few days of future power load forecasting.As the base of making arrangements for power purchase plan and economic distribution of the load,accurate load forecasting is the precondition of security and reliably operation of the power grid.With the improvement of living standards,energy consumption,proportion of attemperation load is increasing,power grid weather sensitive load rising,thus forming the electricity peak load,and widening the peak valley difference.Resulting in when dealing with complex weather conditions the prediction accuracy of existed short-term load forecasting technology is difficult to meet the grid request.To implement the requirements of power grid load delicacy management,to further improve the level of the delicacy of power,and to ensure the safe and stable operation of power grid,the studing of load forecasting model that can reflect the variation of load is necessary to improve the accuracy of short-term load forecasting.Since steping into the eletric big data era,stock of the original operation data of power system rise sharply,load forecasting technology and related fields of science technology,such as meteorological and the economy interinfiltrate.Big data will be the productivity of the power grid in the future.When it turns to short-term load forecasting,excavating of big data deeply,is the integration of macro energy thinking and big data tinking with multi factor meteorological modeling,to realiz the delicacy management power load,and an indispensable part to improve the prediction accuracy of the short-term load forecasting.Based on the power of big data,this paper firstly analyzes the load characteristic under the multi factors of meteorological.From different time dimensions,including the annual,seasonal and daily,analysis the effects of weather on the load.to deal with the low precision of short term load when steping into variable weather,the precision of short-term load forecasting curves can not fit the demand of power grid,this paper proposes a concept of complete sequence of meteorological factors,based on the data mining method the meteorology information granulated is set.The spatial multiple regression and lag model combined with multi strategy sensitivity model is developed to forecast the curves' inflection point under complex weather conditions.On the basis of modified k-means cluster method,the meteorology characteristic day was grasped,and then preliminary prediction curve was get.The paper judged the deformation probability intelligently and took optimized correction necessarily.In order to deal with the abrupt change of weather,a curve correction model based on multi granularity meteorological information matching is proposed.Finally,used daily dynamic data flow to update the modelling parameters to forecast precisely.Finally,the proposed method is used to predict the annual load curve of some areas of south China,the high accuracy proves its practicability,especially fit for the complicated and variable weather condition.