Integrated energy system load time-frequency fusion prediction method based on time-varying correlation

Abstract

The invention belongs to an integrated energy system dispatching technology, aims to fully consider the characteristics of multi-energy loads of an integrated energy system and realize multi-energy load joint prediction, and provides an integrated energy system load time-frequency fusion prediction method based on time-varying correlation, which comprises the following steps of: 1) according to the characteristics of the multi-energy loads, carrying out multi-energy load joint prediction; calculating the correlation between the load and different factors; 2) time domain analysis and prediction: reorganizing one-dimensional load data into a two-dimensional structure according to periodic characteristics of the load data by using a dynamic correlation-based feature reconstruction method, and extracting features together with the input load data through a convolutional neural network (CNN); 3) performing frequency domain analysis and prediction, and capturing a mutual relation between a periodic rule and a frequency domain characteristic in the load data; 4) designing an adaptive weight distribution method; and 5) performing short-term prediction on the multi-energy load of the integrated energy system by using a neural network based on the model. The method is mainly applied to energy system scheduling occasions.

Images