A Load Forecasting Study Based On Hurst Exponent And Fractal Theory

Load forecasting of the power system plays an more and more important role in the operation of the power grid. The more accurate the load forecasting is, the more secure, stable, economical and healthy the power system can become. It is of great value for the development of the national economy. In addition, the load forecasting is significant for power system's planning. Precise load forecasting makes long-term planning, construction, scheduling and management more reasonable and scientific. Power companies can make reasonable arrangements for power generation development plan, avoiding excess generating capacity and power shortage situation. In recent years, the methods for load forecasting occur one after another, and they proved scientific and accurate. However, a variety of methods have their inherent flaws, such as responding times being too long, the impact of factors being not comprehensive enough, only being short-term or long-term load forecasting, and so on. In order to solve such problems, we propose a new prediction method that based on the hurst exponent and fractal theory.Firstly, the traditional forecasting methods are analyzed In this paper, which include the weighted average method, regression forecasting method, least squares method, exponential smoothing method and neural network. Then the various forecasting methods' are compared with the resulting predictions and the actual results, and their advantages and disadvantages are analyzed.Secondly, a new forecasting model based on hurst exponent and fractal theory is established. Historical load data are as the number of disordered columns, their value of hurst exponent are calculated which is an important indicator to judge the long-term trend. The hurst exponent is calculated by the historical load data, in addition the hurst exponent in a certain period of time is calculated through the fractal theory, and then the future loads a certain period using the "squeeze method" are predicted.Finally, it is verified that the method in this paper can not only make the long-term trend of a load area of judgment, but also predict more accurately the value of short-term load through the case analysis. Compared with a large number of experiments results,it shows that the load forecasting method is scientific and accurate, which is of great significance for the load forecast.