“School of Physics”

Back to Papers Home
Back to Papers of School of Physics

Paper   IPM / P / 7116
School of Physics
  Title:   Identification of Hydroxyl Functional Group and Alcohols by Near-Infrared Spectroscopy and Artifical Neural Networks
1.  A. Massoumi
2.  M. Babri
3.  S. Rouhani
  Status:   Preprint
  No.:  0
  Year:  1996
  Supported by:  IPM
Hydroxyl functional group identification has become possible by using near-infrared (NIR) spectroscopy and artificial neural network (ANN), without the use of reagents. In this study, NIR spectra of six compounds was fed to back propagation three-layer neural network as a training set and then, spectra of 33 chemicals were tested by ANN. The results indicated that 97% correct identification was obtained by this system.
With regard to training time and the error of prediction, a systematic approach was used to determine the number of hidden nodes of ANN. Activation of output nodes and number of carbons in a series of aliphatic alcohols are highly correlated. Thus, identification of alcohols without the use of chemical reagents is also possible by this treatment.
The information content of this system (NIR-ANN) for identification of hydroxyl functional group and a series of aliphatic alcohols, by the use of Shannon's equation was calculated.

Download TeX format
back to top
scroll left or right