Format

Send to

Choose Destination
J Environ Manage. 2013 May 15;120:84-92. doi: 10.1016/j.jenvman.2013.01.018. Epub 2013 Mar 19.

Neural network processing of microbial fuel cell signals for the identification of chemicals present in water.

Author information

1
Department of Civil and Environmental Engineering, University of Pittsburgh, Pittsburgh, PA 15261, USA.

Abstract

Biosensing is emerging as an important element of water quality monitoring. This research demonstrated that microbial fuel cell (MFC)-based biosensing can be integrated with artificial neural networks (ANNs) to identify specific chemicals present in water samples. The non-fermentable substrates, acetate and butyrate, induced peak areas (PA) and peak heights (PH) that were generally larger than those caused by the injection of fermentable substrates, glucose and corn starch. The ANN successfully identified peaks associated with these four chemicals under a variety of experimental conditions and for two MFCs that had different levels of sensitivity. ANNs that employ the hyperbolic tangent sigmoid transfer function performed better than those using non-continuous transfer functions. ANNs should be integrated into water quality monitoring efforts for smart biosensing.

PMID:
23507247
DOI:
10.1016/j.jenvman.2013.01.018
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center