Experimental and validation with neural network time series model of microbial fuel cell bio-sensor for phenol detection.

Phenol is one of the most commonly known chemical compound found as a pollutant in the chemical industrial wastewater. This pollutant has potential threat for human health and environment, as it can be easily absorbed by the skin and the mucous. Here, we prepared dual chambered microbial fuel cell (MFC) sensor for the detection of phenol. Varying concentration of phenol (100 mg/l, 250 mg/l, 500 mg/l, and 1000 mg/l) was applied as a substrate to the MFC and their change in output voltage was also measured. After adding 100 mg/l, 250 mg/l, 500 mg/l, and 1000 mg/l of phenol as sole substrate to the MFC, the maximum voltage output was obtained as 360 ± 10 mV, 395 ± 8 mV, 320 ± 7 mV, 350 ± 5 mV respectively. This biosensor was operated using industrial wastewater isolated microbes as a sensing element and phenol was used as a sole substrate. The topologies of ANN were analyzed to get the best model to predict the power output of MFCs and the training algorithms were compared with their convergence rates in training and test results. Time series model was used for regression analysis to predict the future values based on previously observed values. Two types of mathematical modeling i.e. Scaled Conjugate Gradient (SCG) algorithm and Time-series model was used with 44 experimental data with varying phenol concentration and varying synthetic wastewater concentration to optimize the biosensor performance. Both SCG and time series showing the best results with R2 value 0.98802 and 0.99115.