Decision tree method applied to
computerized prediction of ternary intermetallic compounds
computerized prediction of ternary intermetallic compounds
(Institute of Image Processing and Pattern Recognition,
Shanghai Jiaotong University,Shanghai 200030, P.R.China)
Shanghai Jiaotong University,Shanghai 200030, P.R.China)
Abstract: Decision tree method and atomic parameters were used to find the regularities of the formation of ternary intermetallic compounds in alloy systems. The criteria of formation can be expressed by a group of inequalities with two kinds of atomic parameters Zl (number of valence electrons in the atom of constituent element) and Ri/Rj (ratio of the atomic radius of constituent element i and j ) as independent variables. The data of 2238 known ternary alloy systems were used to extract the empirical rules governing the formation of ternary intermetallic compounds, and the facts of ternary compound formation of other 1334 alloy systems were used as samples to test the reliability of the empirical criteria found. The rate of correctness of prediction was found to be nearly 95%. An expert system for ternary intermetallic compound formation was built and some prediction results of the expert system were confirmed.
Key words: ternary intermetallic compound; decision tree; computerized prediction