Integrated energy trading algorithm for source-grid-load-storage energy system based on distributed machine learning

Abstract

The highly integrated source-grid-load-storage energy system has received increasing attention in energy transformation strategies. However, the current static network isomorphism algorithm for distributed machine learning cannot meet the energy exchange needs of the integrated energy system. To better solve the energy loss problem caused by energy trading in the power system, prevent the clean energy loss, and ensure the stable operation of the power system, a distributed dynamic network heterogeneous algorithm is designed on the basis of distributed machine learning. The proposed method uses a dynamic network to balance communication load among servers while solving the hidden state vector errors that cannot be corrected timely due to static network isomorphism. Compared with other methods with a sensitivity of 25%, the sensitivity level of the improved algorithm was above 75%. When the accuracy of other algorithms was 50%, the improved algorithm was above 80%. In the application experiment, the temperature reached 50℃ with the increase of the power. The humidity value always remained above 20. Therefore, the proposed algorithm has superior performance and good application effects, providing new ideas for energy trading in source-grid-load-storage energy systems.