|
Abstract An artificial intelligence method guided by physical laws was developed and combined with traditional thermophysical discipline, using computers to replace the process of exploring laws and proposing formulas. Taking a basic thermophysical parameter, gaseous speed of sound, as an example, in response to the inherent limitations of the truncated virial equation in the high-pressure gaseous region, symbolic regression was used to summarize the high-order virial coefficient terms that characterize multi-molecule interactions. Some universal explicit prediction models were obtained and the root mean square deviation of the predicted gaseous sound-speed for 28 substances by the model can be only 0.29%, with the prediction deviations near the critical pressure significantly lower than those of the existing frontier model.
|
|
Received: 29 October 2024
Published: 03 April 2025
|
|
|
|
2025, Vol. 46 
