Study On Medium And Long-Term Electricity Demand Forecasting Method Based On Industry Classification

Detailed electricity demand forecasting is an important trend in the development of medium and long-term electricity demand forecast. At present the total electricity demand is the most common object of medium and long-term power forecast in our country. The accuracy of predicted results is difficult to further improve, and the information provided is very limited. In this situation, study on medium and long-term electricity demand forecasting method based on industry classification is a topic of great research value, which is also the main point of this paper.Firstly, the domestic and foreign industry classification standards are compared analyzed to design an industry classification result which contains 52 industries and is suitable for medium and long-term electricity demand forecasting;Then, two medium and long-term electricity demand forecasting methods are put forward based on industry classification: the first one is the medium and long-term electricity demand forecasting method based on industry classification and the principle of multilevel forecasting and coordinating, the second one is the medium and long-term electricity demand forecasting method based on industry classification and the curve of industries’ power consumption.In the medium and long-term electricity demand forecasting method based on industry classification and the principle of multilevel forecasting and coordinating, first of all, five basic forecasting models are chosen according to the characteristics of annual electricity demand data to establish an annual power combination forecast model, then the same things are done with quarterly electricity demand data to establish a quarterly power combination forecast model; Then a two dimensional and two level coordinating model is designed based on the principle of multilevel forecasting and coordinating to correct the predicted value, and unify them in both time dimension and industry dimension. This method is formed by improving current mainstream method in the area of medium and long term electricity demand forecasting. This method is suitable for those situations such as lack of manpower, time of emergency, or the data is difficult to obtain, because only history data of industry electricity demand are needed to input.In the medium and long-term electricity demand forecasting method based on industry classification and the curve of industries’ power consumption, a method to analyse the curve of industries’ power consumption is put forward by analogy to the bacterial growth curve; Combined with seven kinds of development form of the ideal curve of industries’ power consumption, a power trend forecasting model is established, and is used to forecast the development tendency of "ideal sequence" in annual power sequence; Combined with 23 outside factors indexes, a multiple linear regression model is established to deal with fitting residual error sequence of the power trend forecasting model, so that the impact of external factors’ change on the development of power industry could be introduced; A quarterly power forecast model based on seasonal index is established to forecast quarterly electricity demand.This method is designed from physical significance, in other words, by considering law of industries’ development. It is suitable for the situations, such as human enough, sufficient time, or easy access to data, because in addition to the history data of industry electricity demand, historical data and predicted data of the 23 external factors indexes are also needed to input.A certain city’s actual data is used to test the two medium and long-term electricity demand forecasting methods based on industry classification proposed by this paper, the results show that the prediction effect of the two methods are both better than the traditional methord. Because of the different applicable conditions of the two methods, they should be chosen according to the specific situation in a practical application.