Carbon potential tracking and prediction method for power system based on physical and data double driving

Abstract

The application provides a kind of carbon potential tracking and prediction method based on physical and data double drive of power system, comprising: constructing time sequence characteristic data set;Constructing source side unit double layer collaborative graph, and generating normalized adjacency matrix;Constructing physical guided source side graph neural network model, thereby outputting predicted source side dynamic carbon emission factor and node carbon injection sequence;High-dimensional input tensor is constructed through space-time graph attention mechanism;Constructing whole network space-time graph neural network model, realizing dynamic tracking of carbon potential;Kirchhoff carbon flow conservation law is introduced as physical regularization term in the process of space-time law deduction, a hybrid loss function containing physical regularization term is constructed, in the back propagation process of whole network space-time graph neural network model training, optimal sentinel mechanism is executed, and finally the dynamic carbon potential prediction result of load side verified by physics is output.The application can improve the accuracy and reliability of carbon potential prediction.

Images