REVEALING SIGNIFICANT CHEMICAL COMPOSITION EFFECTS ON THE SUPER ALLOYS PERFORMANCE BY DATA MINING TECHNIQUES APPLICATION

Анотація

Based on the early developed integrated investigation approach combining mutually consistent computer aided: statistical exploratory techniques, multiple regression models development, Monte-Carlo simulations and multipurpose manufacturing technology parameters optimization, computer statistical investigations of chemical composition effects on some basic performance characteristics of modern industrial superalloys (SAs) were conducted. According to the developed approach, the Data Mining-based C&RT statistical technology was used in the current work as the first stage of the further analysis. An important feature of the developed approach to the “passive experimentation” in general and the applied technique in particular was emphasized as both providing minimal investigation expanses, as compared to the “active”  one and so having no alternative in free market economic conditions. As the stage primary results, the dendrograms were constructed providing the following statistically grounded visual information for each SAs controlled performance index: – a full list of statistically significant chemical elements;  allowable variants of the elements combinations; limits of the chemical element concentration intervals spillover of which provides changes in the chemical elements combinations; variation tendencies for the control indexes within the revealed concentration intervals. Interactions of the most of the chemical elements comprised by SAs in their effects on the investigated performance indexes were shown. High effectiveness of the obtained dendrograms application to reveal individual and mutual chemical element effects on the multi element alloys performance together with uniqueness of such an information for industrial alloys were emphasized. Some prospect possibilities were noted for the obtained dendrograms application: - as a first stage of a multi-parameter process analysis in industrial conditions, – a ground stage of  the multiple regression models further development and final multipurpose technological processes optimization conducting.