Artificial Neural Network‐Based Quantitative Structural Property Relationship for Predicting Boiling Points of Refrigerants

Quantitative Structure‐Property Relationship (QSPR) model for the estimation of boiling points of organic compounds containing halogens, oxygen, or sulfur without hydrogen bonding were established with the Molecular Modeling Pro Plus (MMPP) software. A QSPR study was performed to develop models that relate the structures of 90 refrigerants compounds to their boiling point temperatures. Molecular descriptors derived solely from structure were used to represent molecular structures. A subset of the calculated descriptors selected using genetic algorithm (GA) was used in the QSPR models development. Artificial neural network (ANN) is utilized to construct the QSPR model. The optimal QSPR model was developed based on a 4‐4‐1 artificial neural network architecture using molecular descriptors calculated from molecular structure alone. The root mean square errors (RMSE) in normal boiling points predictions were 4.46 °C for the training set, 3.86 °C for the validation set and 4.99 °C for the prediction set. The prediction results are in good agreement with the experimental value.