Search

[Display omitted] • Two-stage models were developed to estimate biogas upgrading in MECs excellently. • NARX-BP hybrid neural network was proposed to predict CH 4 production via DET. • NARX model presented perfect preference of predicting CH 4 , TIC and CO 2 yield. • ANNs achieved high level among different machine learning methods in modeling. • Various machine learning models and ANNs topologies were implemented and assessed. With the increasing of data in wastewater treatment, data-driven machine learning models are useful for modeling biological processes and complex reactions. However, few data-driven models have been developed for simulating the microbial electrolysis cells (MECs) and traditional models are too ambiguous to comprehend the mechanisms. In this study, a new general data-driven two-stage model was firstly developed to predict CH 4 production from in-situ biogas upgrading in the biocathode MECs via direct electron transfer (DET), named NARX-BP hybrid neural networks. Compared with traditional one-stage model, the model could well predict methane production via DET with excellent performance (all R2 and MES of 0.918 and 6.52 × 10−2, respectively) and reveal the mechanisms of biogas upgrading, for the new systematical modeling approach could improve the versatility and applicability by inputting significant intermediate variables. In addition, the model is generally available to support long-term prediction and optimal operation for anaerobic digestion or complex MEC systems. [ABSTRACT FROM AUTHOR]