Artificial Neural Networks for the Automated Detection of Trichloroethylene by Passive Fourier Transform Infrared Spectrometry

Abstract

Artificial neural networks are applied to the automated classification of trichloroethylene (TCE) signatures from passive Fourier transform infrared remote sensing interferogram data. Through the use of three data collection methods, a combination of laboratory and field data is acquired that allows the methodology to be evaluated under a variety of infrared background conditions and in the presence of potentially interfering compounds such as sulfur hexafluoride, methyl ethyl ketone, acetone, carbon tetrachloride, and ammonia. To maximize the computational efficiency of the network optimization, experimental design techniques are employed to develop a training protocol for the network that takes into account the relationships among five variables that are related either to the network architecture or to the training process. This protocol is implemented for the case of a back-propagation neural network (BNN) and is used to develop an optimized network for the detection of TCE. The classification performance of the network is assessed by comparing both TCE detection capabilities and false detection rates to similar classification results obtained with the technique of piecewise linear discriminant analysis (PLDA). When applied to prediction data withheld from the optimization of both the BNN and PLDA algorithms, the BNN method is observed to outperform PLDA overall, with TCE detection rates in excess of 99% and false detection rates less than 0.5%.

Reliable component identification in atmospheric open‐path FTIR spectroscopy by a cross‐correlation method

“…In Reference 10, two-and three-component mixtures were tested with the use of library spectra that were preprocessed by a Fourier transformation and a subsequent orthonormalization based on a principal-component analysis. Intensive computation is also needed within the training of neural networks, [2][3][4] in particular, when a large number of different compounds must be detected. Compound identification by our straightforward criterion can be implemented without intensive computer programming, but appropriate and component-characteristic spectral intervals have to be selected for routine work.…”

“…The latter technique has been employed for qualitative assays with the use of pattern-recognition techniques (e.g., digital filtering and piecewise-linear discriminant analysis techniques 1 or artificial neural network approaches. [2][3][4] The potential for chemical-warfare-agent detection has prompted the military community to contribute to the development of this technique.…”

Automated Detection of Chemical Vapors by Pattern Recognition Analysis of Passive Multispectral Infrared Remote Sensing Imaging Data

“…Previous work in our laboratory on passive infrared rem ote sensing has focused primarily on chemical vapor detection from nonimaging hyperspectral sensors positioned on the ground. 4,9,10 In the work described here, autom ated detection algorithms are developed for use with multispectral imaging measurem ents m ade from an aircraft platform. The system is used to acquire two-dimensional images of a nitrogen fertilizer facility across 14 infrared bands.…”

Multiple Filtering Strategy for the Automated Detection of Ethanol by Passive Fourier Transform Infrared Spectrometry

“…Work in our laboratories has focused on increasing the viability of passive FT-IR remote sensing m easurements through the development of data analysis algorithms that help to compensate for the data variance associated with changes in the background radiance. [6][7][8][9] The strategy used is to apply digital ltering techniques to the raw interferogram data collected by the FT-IR spectrometer. This approach takes advantage of the fact that the background radiance is primarily represented as a broad spectral feature whose representation in the interferogram is constrained to the region near the point of zero path difference (ZPD).…”