The Research Of Load Forecasting Method Based On Regional Power Supply Enterprise

The accuracy of power load forecasting has become one of the symbols to measure the power system operation management modernization. The correct power system scheduling, planning and operation are inseparable from the electric power load forecasting. Accurate load forecasting plays a very important role in the security, reliability, and economic operation of power system. In this paper, the author analyses the characteristics and applicable scope of the various methods of power load forecast firstly, and then on the basis of the study of neural network and fuzzy theory, the author proposes a short term load forecasting model which combined Elman neural network and fuzzy theory together.The new method analyses the historical load data in horizontal and vertical way. The author investigated the input variables of neural network which is the key problem to influence the prediction effect. because it is important to select forecasting model. The dimension of the input variables impacts the structure of the neural network seriously. In this paper, the author applied the concept of autocorrelation function on the selection of input variables of the neural network modal. Through analysis on the characteristics of historical load, the author combined autocorrelation function theory and coscodl small principle to select input variables of neural network forecasting model. BP neural network, RBF neural network and Elman neural network are all used to establish a short-term daily load forecasting model. Through a lot of simulation research and a contrast between the results and the prediction results, the author combines Elman neural network with fuzzy theory to establish a combination forecasting model. The author used the simulation software to do a short-term load forecasting test in the actual regional power supply enterprise., The results shows that compared with the single neural network model, the forecasting method proposed in this paper has improved the forecast accuracy to some extent, especially in the problem of load forecasting for the weekend and holidays.