Power load prediction method based on dynamic decomposition-reconstruction integrated processing

The invention discloses a power load prediction method based on dynamic decomposition-reconstruction integrated processing, and the method comprises the steps: firstly obtaining power load data, carrying out the preprocessing, carrying out the decomposition of the data through mlptdense decomposition, building a GCN-Reformer power load prediction model according to the decomposed components, carrying out the optimization of the Reformer hyper-parameters through an improved OOA algorithm, and carrying out the prediction of the power load. Selecting low-precision components needing secondary decomposition according to performance evaluation of decomposed components on a verification set, aggregating all the low-precision components by adopting permutation entropy to obtain a high-complexity component and a low-complexity component, performing secondary decomposition by adopting a WPD method, and then predicting all WPD decomposed D components by using a GCN-Reformer model, whether decomposition needs to be carr