Power Short-term Load Forecasting Based On Fractal Theory

Load prediction is a important part of the construction of electricity transmit and transforming, short term toad forecasting technique directly affects the performance of generator in power plant and grid attempter.According to the development status of load forecasting technique and the actual needs of project, aiming at the deficiencies of existing similar historical day load forecasting method, this thesis carries out a deeply study to the load forecasting technique based on fractal geometry theory, and develops a load forecasting method which suits the effective demand of project.When data preprocessing is performed, vertical and horizontal pretreatment is used in the paper to preprocess historical load data which establish the basis for short term load forecasting model using these historical data; Secondly, fractal and time serial characteristic of load data is used to designed a fractal dimension calculate method for disperse sampling data; Then, in viewing of the insufficiency of present shortage of arithmetic to traditional similar historical day forecasting method used inverse operation of sand boxing dimension to get the load of next day, and find why error appears; Finally, makes the forecasting data image by fractal interpolation, vertical proportion parameter used in the interpolation was determined by the similar historical load data, the image can review change condition between load spot.The morbid data distinguish algorithm based on fractal characteristic, electrical load sampling data fractal dimension algorithm, load forecasting algorithm based on inverse operation of sand boxing dimension and vertical proportion parameter algorithm in fractal interpolation proposed by this article, can solves the traditional methods' insufficiency. So this algorithm has certain application value in the short term load forecasting.