Verification of SCO2 Recompressed Brayton model of EBSILON software based on BP neural network

The recompression structure in SCO
2
Brayton cycle is the most simple and efficient model, and most literature has adopted this structure. EBSILON software is a thermal system modeling software, which can be used for simulation modeling of SCO
2
cycle. In order to verify the effectiveness and correctness of the SCO
2
recompression Brayton model built by EBSILON software, a prediction model is built using BP neural network to validate the simulation data. The R2 and MAE of the training set are 0.99421 and 0.16043, and the R2 and MAE of the test set are 0.99658 and 0.174, respectively. The errors are within the allowable range, and the predicted value is basically consistent with the real value, which indicates that BP neural network can more accurately prove the effectiveness of the SCO
2
recompressed Brayton model established by EBSILON software, and this software can be used to continue modeling and simulation.