Medium-and-long-term Forecasting Of Beijing Electricity Demand Based On Composite Knowledge Mining

Electricity demand forecasting is one of the important tasks in the power sector, which is not only the premise of the reasonable power macro planning, but also the important guarantee for the safe and economic operation of power system. How to achieve accurate and effective medium and long term electricity power demand forecasting is one of the key and difficult points for the electricity power department. In this paper, we take medium-and-long-term forecasting of Beijing electricity demand as an example, building a power demand forecasting model based on complex knowledge mining by making knowledge mining on various quantitative and text factors influencing Beijing city electricity demand.This paper first introduced the theory of power demand forecasting, including the concept, characteristic, classification and influence factors of electricity power demand forecasting, and according to their own characteristics, these influence factors were divided into quantitative factors and text factors; The main methods of power demand forecasting were analyzed which were divided into four kinds of methods named empirical methods, classical methods, statistical forecasting methods and intelligent forecasting methods; Secondly, this paper introduced the concept of compound knowledge mining and its two core technology of data mining and text mining from three aspects:concept, procedure and methods. A set of quantitative factors that affected the power demand of Beijing city were made and two kinds of data mining technology namely correlation analysis and Grainger causality test were used to make data mining on these factors in order to dig out the main quantitative factors influencing electricity demand of Beijing City; A set of text factors affecting the power demand of Beijing City were given, and the FP-Tree association rule algorithm was used to dig out the text factors which had an very important impact on Beijing electricity demand; On the above research, a mid and long-term power demand forecasting model based on complex knowledge mining for Beijing city was established by RBF neural network model and decision tree technology which combined the results of data mining and text mining; And last, the model was used to forecast the mid and long-term power demand in Beijing city.The paper made knowledge mining on various quantitative and text factors influencing the medium and long term electricity demand of Beijing city and built a power demand forecasting model of Beijing based on complex knowledge mining creativity. The research of this paper has some reference value for achieving more accurate and effective power demand forecasting and enriching existing power demand forecasting theory.