Power load prediction method and system

Abstract

The invention provides a power load prediction method and system, and the method comprises the steps: dividing a historical load time sequence into a plurality of historical DTBs according to a non-uniform time window, carrying out the nonlinear transformation of each historical DTB through Logistic mapping, carrying out the multi-scale fractal feature extraction of the DTBs, carrying out the first prediction based on the fractal features, and obtaining the first prediction load of each DTB in the future, and calculating a first prediction residual error of each DTB in the history, constructing a space-time correlation network, performing second prediction to obtain a second prediction load of each DTB in the future, calculating a second prediction residual error of each DTB in the history, and performing fuzzy fusion on the first prediction load and the second prediction load based on a residual error subjective sum of two times of prediction to obtain a final prediction load. According to the method, dual prediction is utilized, secondary load prediction is carried out by capturing the internal law of the load data and the mutual relation in time, and the prediction results are effectively fused, so that the advantages of various predictions are combined, and the load prediction accuracy and reliability are improved.

Images