ARTIFICIAL NEURAL NETWORK OPTIMIZATION USING GA AND PSO ALGORITHMS TO PREDICT REFRACTIVE INDEX OF BINARYSOLUTIONS

In this study, an artificial neural network was used to predict the refractive index of the binary solutions, including alcohol-alkane, alkane-alkane and alcohol-alcohol. Inputs of the neural network are molecular mass of the first substance, second substance molecular weight, temperature, mole fraction of the first substance, functional groups, and the output of neural network is refractive index of binary solutions. Optimal design for a neural network, is a network with the error back propagation algorithm with Levenberg Marquardt training function and tangent hyperbolic transfer function for hidden layer and linear transfer function for the output layer. Neural network modeled has 8 inputs, 10 neurons in the hidden layer, and one neuron in the output layer. Then, we optimized the neural network first using genetic algorithm, after that using particle swarm optimization algorithm. The results show that the f the refractive index can be predicted using ANN-GA model with a 2.4% error and using ANN-PSO model, with error of 2.235%. The results show that ANN-PSO model has higher ability to predict and interpolation refractive index of binary solutions data, including alcoholalkane, alkane-alkane and alcohol-alcohol compared to the ANN-GA model.