Prediction of air specific heat ratios at elevated pressures using a novel modeling approach

Abstract

An accurate and efficient model based on least-squares support vector machines (LSSVM) is developed for the determination of air specific heat ratios at elevated pressures. Additionally, the coupled simulated annealing optimization strategy is used to calculate the optimal values of the LSSVM parameters. A large dataset of air specific heat ratios as a function of temperature and pressure for about 170 samples is used to develop and validate the model. The leverage approach (Williams plot) is used to determine the applicability domain of the model and to detect probably erroneous data points. Comparison of the obtained results with a previously published correlation as well as an intelligent method demonstrates that the performance of the presented model is more satisfactory than that of other methods. An accurate and efficient model based on least-squares support vector machines is developed for the determination of the air specific heat ratio at elevated pressures. The proposed model is able to predict the specific heat ratio with an accuracy adequate for many practical applications. The method is compared to previously published correlations and demonstrates better performance.