Abstract： The determination of types and size distributions of inclusions in steel samples from Optical Emission Spectrometry with Pulse Discrimination Analysis (OES/PDA) data requires advanced mathematical tools. The main task is to filter out those OES/PDA events that correspond to inclusions in contrast to volume events, i.e. to find reliably the outliers in the data sets. The distribution functions of intensities and even more of the masses are non-normal. Thus the data are fitted to non-normal distributions and the outliers are determined as events not belonging to these distributions. After determining the outliers this information is used to find clusters of data points corresponding to distinct inclusion types. For this purpose a particular kind of neural network called Self Organizing Map (SOM) proved to be well suited. With the information about outliers and inclusions types, mass distributions and further size distributions of the different inclusion types can be calculated.
［1］ Meilland R, Dosdat L. Rapid characterization of inclusionary cleanliness in steels by PDA-OES mapping［J］. Revue de Métallurgie. Paris, 2003, 4：373-382.
［2］ Pande M M, Guo M, Dumarey R, et al. Determination of steel cleanliness in ultra low carbon steel by pulse discrimination analysis-optical emission spectroscopy technique［J］. ISIJ International, 2011,51 (11): 1778-1787.
［3］ Shin Y, Bae J S. Statistical signal processing of optical emission spectroscopy for micro inclusion detection in steel［C］. Proc. of IEEE Instrumentation and Measurement Tech. Conf., Budapest, Hungary, 2001, 2: 1089-1092.
［4］ Rousseeuw P J, Van Driessen K. A fast algorithm for the minimum covariance determinant estimator［J］. Technometrics, 1999, 41(3), 212-223.
［5］ Fletcher R. Practical methods of optimization (2nd ed.) ［M］ New York: John Wiley & Sons, 1987.
［6］ Hestenes M R, Stiefel E. Methods of conjugate gradients for solving linear systems［J］. Journal of Research of the National Bureau of Standards,1952, 49 (6): 409-436.
［7］ McLachlan G, Krishnan T. The EM Algorithm and Extensions［M］. New York: John Wiley Sons, 1997.
［8］ Breiman L, Friedman J H, Olshen R A, et al. Classification and Regression Trees［M］. Monterey, CA: Wadsworth & Brooks/Cole Advanced Books & Software, 1984.
［9］ Haykin S. Neural Networks: A Comprehensive Foundation［M］. Prentice Hall, 1998.
［10］ Kohonen T, Learning Vector Quantization. In: Arbib M. A. ed. The Handbook of Brain Theory and Neural Networks ［M］. Cambridge, MA, MIT Press, 1995, 537-540.
［11］ Kohonen T. Self-Organising Maps［M］. Springer, Berlin, 1997.
［12］ Hartigan J A. Clustering Algorithms［M］. Wiley, 1975.
