The application of Kohonen neural networks to diagnose calibration problems in atomic absorption spectrometry

In atomic absorption spectrometric measurements calibration lines are measu red daily. These lines are not always acceptable. They can, for instance, c ontain outliers, have a bad precision or can be curved. To evaluate the qua lity of those lines a method which gives a fast diagnosis is recommended. I n this study the use of Kohonen neural networks was examined as an automate d procedure to classify these calibration lines. The results were compared with those obtained using a decision support system which uses classical st atistical methods to classify the lines. The prediction capabilities of bot h approaches relative to a visual inspection and classification was found t o be comparable, or even slightly better for the Kohonen networks, dependin g on the training set used. For both techniques a prediction error rate of < 10% was obtained, relative to a visual classification. (C) 2000 Elsevier Science B.V. All rights reserved.