Predicting the heats of combustion of polynitro arene, polynitro heteroarene, acyclic and cyclic nitramine, nitrate ester and nitroaliphatic compounds using bee algorithm and adaptive neuro-fuzzy inference system

Abstract

A new method was developed for prediction of the heats of combustion of important classes of energetic compounds including polynitro arene, polynitro heteroarene, acyclic and cyclic nitramine, nitrate ester and nitroaliphatic compounds. A set of 1497 zero- to three-dimensional descriptors was generated for each molecule in the data set. A major problem of modeling is the high dimensionality of the descriptor space; therefore, descriptor selection is one of the most important steps. In this paper, bee algorithm (BA) was used to select the best descriptors. Bee algorithm is a new population-based optimization algorithm, which is derived from the observation of real bees and proposed to feature selection. Four descriptors were selected and used as inputs for adaptive neuro-fuzzy inference system (ANFIS). Squared correlations of coefficients were obtained as 0.9980, 0.9996 and 0.9988 for training, test and validation sets, respectively. In comparison with genetic algorithm (GA)-ANFIS and multiple linear regression (MLR)-ANFIS, the results showed that Bee-ANFIS can be used as a powerful model for prediction of heats of combustion of these compounds.