Monthly electricity consumption forecast of the park based on hybrid forecasting method

Abstract

Along with the forecast of electricity consumption can provide the basis for the power supply enterprise to control the electricity sale market, at the same time, it can make scientific planning and guidance to the generating capacity of the electric power company, in the future development and planning of distribution network, under the the new electric power reform,the park has become an important experimental area for electric power reform, electricity consumption forecast are beginning to orient the power needs of small-scale users, due to the randomness of electricity consumption of small-scale users is large, it has great effects on the prediction results, the single exponential smoothing can only reflect the overall change of monthly electricity consumption in the park, fail to reflect the fluctuation characteristics of electricity consumption with seasonal changes.Therefore, according to the temporal characteristics of electricity consumption data, this paper combiness the design ideas of time series method and exponential smoothing method, optimizes the exponential smoothing model, introduces the time series model, and establishes a new improved model.Firstly, the seasonal decomposition model is used to carry out personalized decomposition of the electricity consumption sequence of the corresponding month, and the electricity consumption sequence is decomposed into trend component, seasonal component and random component, so as to avoid mutual interference between the predicted components.And then choose the appropriate exponential prediction model and time series models to forecast the component,this model makes use of the basic principle that the recent data has large influence on the prediction and the long-term data has little influence on the prediction, considering the characteristics of the three components change over time, using a variety of model fitting prediction, the method excludes alpha and beta parameters, the gamma smooth parameter controls the exponential decline of the seasonal component,the larger the value is , the greater the closer the observed value is to the seasonal effect weight is.The improvement of the forecast accuracy of electricity consumption can effectively reduce the cost of power generation, improve the economic and social benefits, and promote the planning and development of the distribution network in the future. In this paper, the algorithm is compiled based on R language, and the validity of the proposed method is verified and analyzed based on the actual monthly electricity consumption data of the park.The results show that this method can improve the accuracy of prediction.