Ultra Short Term Forecasting Of Power Load Based On Time Series Method

With the rapid development of economy,the electric power sector is developing rapidly under the influence of the economy,and the construction of power network is constantly improving.The power sector has become a pillar industry of modern society,as an indispensable power in modern life resources,to provide enough energy for various industries,so that each link can be carried out smoothly,the power department of rational allocation of resources is particularly important.Therefore,the quality and reliability of power supply is of great significance to the sustained and stable development of the national economy.In this case,the ultra short term load forecasting is paid more and more attention by the electric power sector,and how to improve the accuracy of prediction is the key problem in the premise of sufficient power.The improvement of prediction accuracy is helpful to the maintenance of power system,which can help the dispatchers to analyze the adequacy of the spinning reserve capacity.Thus,when the power reaches the peak state,the reasonable allocation can be realized,and the coordination of the hydro power generation and the thermal power generation can be realized,so as to avoid unnecessary losses.The main purpose of this paper is to predict the value of ultra short term power load.With the method of time series based,first of all,a simple description of the present situation of research on ultra short term load forecasting at home and abroad,in order to have an overall grasp of the development level of the ultra short term load forecasting;then,the classification and characteristics of load forecasting are introduced,and extends to the ultra short term load forecasting and elaborated the scientific prediction and significance to study the role of the corresponding to the ultra short term load,and introduces the characteristics of ultra short term load forecasting and its significance;secondly,in the third part of the time series models including AR,MA,ARMA and ARIMA model definition,basic properties and methods of the tool in detail the introduction,and the modeling steps list,respectively on time series stationarity test,time series smoothing processing,model order and parameters of the model Estimation and model optimization;in the fourth part,based on the theoretical basis of the first three chapters,this paper selects the Belgian region from January 1,2017 to January 5th day 24 h with 1H interval of 120 groups of real time power total load data,stationarity test data,using differential operation for processing,to achieve data smoothing.In thispaper,the first order difference and the two order difference are carried out respectively,and then the model is established,and the AIC model is used to select the better model to predict the data more accurately.After selecting the optimal model,the 24 sets of real-time load data of 24 h are used to make the short-term load forecasting.In the fifth part,by comparing the predicted results with the real values,the relative errors between the predicted results and the real values are analyzed in detail.