Application of T-S neural network in prediction for anaerobic bioreactors

Key words: 
• fuzzy neural network /
• toxic load /
• anaerobic bioreactor /
• prediction

Abstract: After the three bioreactors became steady-state, they were shocked by the chloroform and 2, 4-dinitrophenol. The T-S fuzzy neural networks were created based on the database collected from the anaerobic system shock, and were used to predict the biogas production rate, volatile fatty acid and CH4 of the bioreactors. The results showed that the fuzzy neural network based on a bioreactor can perfectly predict performance of the bioreator, its correlation coefficients of observed and simulated values were above 0.850 for both training data set and testing data set. But the fuzzy neural network based on a bioreator could not predict well the other bioreactor, the values of correlation coefficients of observed and simulated were below 0.500 for the biogas production rate, volatile fatty acid and CH4.