Improvement of artificial neural network for prediction of cell voltage and flow rate in a membrane cell by genetic algorithm

Familiarity with the internal structure and parameters of artificial neural networks has important and considerable effects on prediction of efficiency and return. Although there are some guides for selection of geometric and internal parameters in the previous researches, but most of networks have been calibrated using test and trial method. This paper shows use of genetic algorithms (Gas) for improvement of back propagation neural network (BPNN) weights including various layers. Prediction of functional efficiency of ANN internal structures improvement by GA using an experimental database published from cell performance. As specified, ANN network improved by GA had the efficiency and return exacter than a network therein ANN calibration has been performed using a test and trial method.