Artificial neural networks applied for studying metallic complexes

Metallic complexes of multimetal and multiligand systems are complicated fo r calculating equilibrium concentrations in solutions. An artificial neural network has been developed for studying Al3+ and EDTA complexes in solutio n with an initial concentration of 0.01 mol L-1 for these species. In this system there are 20 compounds and may exist 18 simultaneous reactions. The neural network has been trained and the simulated data of different concent rations as a function of PH are predicted with an accuracy of about 1% for all species simultaneously. A general analytical formula is presented, whic h directly relates all the concentrations as a function of pH. The analysis showed that predictions closer to the boundary of the input and output dat a are quantitative while out of these limits these are not even qualitative . (C) 2001 John Wiley & Sons, Inc.