Data Mining Technology In The Metallurgical, Chemical Applications

Tata Mining and knowledge Discovery" is the technique to use computer and its software to extract useful information from complicated data sets. This thesis is dealing with some methods and some topics of application of data mining in the fields of metallurgy, materials sciences and chemical engineering. 1. K-Nearest Neighbor (KNN) method, one of commonly used pattern reoognition classification method, has been modified for the purpose of feature selection, practice of computation indicates that this method is more suitable than many traditional methods of feature selection when data structure is inclusion type. 2. An extend Miedema's model has been used to study the regularities of formation of ternary intermetallic compounds. Four atomic parameters (φ,Nws1/3,R,Z ) or their functions are used as the features to span the multi-dimensional spaces, and the representative points of temaxy systems are plotted into these spaces. Based on the regularities found, an expert system has been built for the prediction of formation of ternary intermetallic compounds. This expert system has been used to "predict"the formation of 78 ternary compounds in newly published phase diagrams of ternary alloy systems and results are rather good. 3. Data Mining methods are used to investigate following practical problems: (1) The data of corrosion of automobile parts of Ford Company have been analyzed with a mathematical model found by our data mining software. (2) The data of springback of fabrication of steel plate in Baoshan Iron Steel Company have been investigated by data mining method. (3) The data of testing of DNA chips have been investigated by data mining method. (4) The data of the cost of repair of copying machines of a US Company have been investigated by data mining method.