Evaluation of artificial neural network algorithms for predicting the effect of the urine flow rate on the power performance of microbial fuel cells

Abstract

Microbial fuel cell (MFC) power performance strongly depends on the biofilm growth, which in turn is affected by the feed flow rate. In this work, an artificial neural network (ANN) approach has been used to simulate the effect of the flow rate on the power output by ceramic MFCs fed with neat human urine. To this aim, three different second-order algorithms were used to train our network and then compared in terms of prediction accuracy and convergence time: Quasi-Newton, Levenberg-Marquardt, and Conjugate Gradient. The results showed that the three training algorithms were able to accurately simulate power production. Amongst all of them, the Levenberg-Marquardt was the one that presented the highest accuracy (R = 95%) and the fastest convergence (7.8 s). These results show that ANNs are useful and reliable tools for predicting energy harvesting from ceramic-MFCs under changeable flow rate conditions, which will facilitate the practical deployment of this technology.

Data-driven modelling for resource recovery: Data volume, variability, and visualisation for an industrial bioprocess

“…The model data statistics and their bootstrap 95% confidence intervals in square brackets [dissolved oxygen: DO; conductivity: Cond. ; chemical oxygen demand: COD; total suspended solids: TSS; total organic carbon: TOC; ammonium: NH data is lower than some ANN models in the literature (R 2 : 0.96 [52], R 2 : 0.95 [53] and R 2 : 0.99 [54]). The higher accuracy reported by these models is likely caused by the models being developed from data that was collected under laboratory conditions, therefore, not subject to the same degree of variability exhibited by the data collected from the H 2 AD plant.…”

Post-COVID-19 pandemic and the Paris agreement: a socioeconomic analysis and carbon emissions forecasting in developed and developing countries

“…The general output and error function of ANN can be expressed in Eqs. ( 3 ) and ( 4 ) (de Ramón-Fernández et al 2020 ). where, X i and W ij represent the input data and a weight value, respectively, f () gives the activation function, and y i and D i are the net output and the estimated output value, respectively.…”

“…The general output and error function of ANN can be expressed in Eqs. ( 3) and ( 4) (de Ramón-Fernández et al 2020).…”

Establishing a Novel Algorithm for Highly Responsive Storage Space Allocation Based on NAR and Improved NSGA‐III

“…Stage 2 involved the error testing of predicted results. e algorithm evaluated the prediction effect using the RMSE, MSE, and coefficients of autocorrelated errors [59][60][61]. If the error in prediction was within the acceptable range, the predicted results of the data used as inputs for the next stage of storage space allocation.…”