Estimation of LLE Data for Binary Systems of N-Formylmorpholine with Alkanes Using Artificial Neural Network–Genetic Algorithm (ANN–GA) Model

Abstract

The purpose of this work was to predict liquid-liquid equilibrium of binary systems including N-formylmorpholine (NFM) with alkanes (heptane, nonane, and 2,2,4-trimethylpentane) over the temperature range from around 300 K to 420 K. Therefore, three feed-forward artificial neural network (ANN) models were developed for the three systems. Compositions of alkanesin light phase and heavy phase were considered as network inputs, and the temperature was the output variable. Genetic algorithm (GA) method was used to design the neural network. It minimized the total mean squared error (MSE) between net output and desired output with optimizing weights and biases of the ANN. The validity of the models was evaluated through a test data set, which was not used in the training data set. The results of this work show that the hybrid of artificial neural network and genetic algorithm (ANN–GA) can estimate the LLE of the binary systems with high precision.