Prediction of flame ionization detector response factors using an artificial neural network

An artificial neural network (ANN) was successfully developed for the modeling of flame ionization detector response factors. The generated ANN was evaluated and applied for the prediction of response factors of several varieties of organic compounds. The results obtained using neural network were compared with different sets of experimental values as well as with those obtained using multiple linear regression technique. Comparison of neural network standard error of prediction values with those obtained using regression equations shows the superiority of ANNs over that of regression models. Calculations of Dietz response factor for two different prediction sets show that an ANN has a good predictive power. (C) 1998 Elsevier Science B.V. All rights reserved.