Estimating the Fates of C and N in Various Anaerobic Codigestions of Manure and Lignocellulosic Biomass Based on Artificial Neural Networks

In this study, artificial-neural-network- (ANN-) based models were explored and validated to predict the fates of carbon (C) and nitrogen (N) under 84 types of digesters in treating different blends of seven substrates (corn straw, rice straw, wheat straw, swine manure with fed feedstuff and foodstuff, cattle manure, and chicken manure) under changing volatile solids (VS) loadings. ANN models based on principal component analysis (PC-ANN) were developed for estimating the fate of C (CH4 yields and COD concentrations in the supernatant), showing higher prediction accuracies than the original ANN models. The fate of N (NH4+-N concentrations in the supernatant) was well predicted by the ANN model with two inputs, namely, total Kjeldahl nitrogen (TKN) and total ammonium nitrogen (TAN) in the substrates. The models were also developed for wide applications to validate the CH4 yields and NH4+-N concentrations for new databases outside the established data range obtained from the literature, with regression coefficient (R2) values of 0.705 and 0.791, respectively. This study can provide guidance for future process optimization and nutrient recycling in anaerobic digestion (AD).