THE USE OF ARTIFICIAL NEURAL NETWORKS IN PLASMA SPECTROSCOPY

Plasma spectroscopic instrumentation produces a vast amount of data th at requires a substantial effort in analysis. New methods are needed t o improve the efficiency of the analysis process. Artificial neural ne tworks, which have been applied to a variety of signal processing and pattern recognition problems, are well suited for the analysis of labo ratory data that can be viewed as a pattern matching problem. We summa rize the neural network concepts and algorithms and illustrate the use of a neural network in the determination of electron temperature and density by analysis of spectral data from a plasma.