Study On Monthly Electricity Sales Forecast Based On Season Adjusted ARIMA Model

Regional monthly electricity sales forecast is an important basis for the planning,construction and economic operation of regional power grids.Accurately forecasting the monthly electricity sales in the region and the regulation of electricity consumption can effectively improve the utilization rate of equipment and optimize the electricity consumption mode,and can further refine and improve the electricity market Power supply business services to meet the needs of different users of electricity.However,the traditional method is limited to the time-series trend extrapolation in the monthly sales forecast,without considering the influence of various internal and external factors on the sales forecast,which will lead to the loss of data information in the sales forecasting and the forecast error Increase.Based on the real-time data of the sales information,based on the ARIMA model and the seasonal adjustment function in the EVIEWS software,the monthly sales of electricity is decomposed based on the traditional monthly sales analysis method,The ARIMA model predicts monthly electricity sales in Beijing based on monthly electricity sales data from 2008 to 2015 in Beijing,providing more accurate and timely data support for Beijing’s electricity supply and marketing management.Specifically,this paper first defines the concept of sales volume forecast and season adjustment,analyzes the basic principles and steps of sales volume forecast,the classification of sales volume forecast and the reasons for the sales forecast error.Secondly,this paper analyzes and forecasts the current situation of Beijing’s electricity sales from the perspectives of sales of electricity in the whole society,electricity sales in various industries,electricity sales in major industries with high energy consumption,and sales volume in sub-regions,and analyzes the current situation of sales in Beijing from five dimensions Power problems,and stressed the importance of research on the sale of electricity.Thirdly,the monthly electricity sales data of Beijing from January 2008 to December 2015 are taken as samples,and the monthly sales of Beijing are predicted by three traditional methods: linear regression,Holt-Winters model and ARIMA model respectively.Compared with the traditional methods,the author points out the insufficiency of the traditional model prediction methods and the improvement direction.Fourthly,based on the analysis of the influencing factors ofBeijing monthly electricity sales through regression analysis,this paper uses EVIEWS software to decompose the monthly electricity sales data through the seasonal adjustment function provided by X12 multiplication model in EVIEWS software,Improve the traditional ARIMA model,forecast the monthly electricity sales in Beijing.Finally,this paper uses the data from January 2008 to December 2013 in Beijing as the test sample and monthly electricity sales data from January 2014 to December 2015 as the training samples to verify the ARIMA model Based on which the monthly electricity sales of Beijing from January to December 2017 are predicted,and then analysis the results of the prediction to get the conclusion of the research.