NEURAL NETWORK SYSTEM USING SPECTRAL ANALYSIS FOR IDENTIFICATION OF COMPLEX ORGANIC STRUCTURES

摘要

Apparatus and methods for recognizing and identifying materials. Characteristics are obtained spectra of the materials using spectroscopic techniques including spectroscopy, nuclear magnetic resonance, infrared absorption analysis, X-ray analysis, mass spectroscopy and the gas chromatography. Desired parts of the spectra can be selected and then placed in a form and in an appropriate format for presentation to a number of input layer neurons in a neural network independent. The network is firstly formed at a predetermined formation process, it can then be used to identify particular materials. Said apparatus and said methods are particularly useful to recognize and identify organic compounds such as complex carbohydrates, whose spectra require a high level of training and several hours of detailed work to allow identification and are often indistinguishable from each other by a human interpretation.