Application of Regression Trees in Optimization of Metal Forming Process

Abstract

Application of sensitivity analysis in optimization of process parameters of production processes for innovative materials, e.g. dual phase steel, requires deterministic model of thermomechanical processes and large datasets that covers whole surface of results. Difficulties in optimization of process parameters correspond with large number of control variables, which should be considered in the technology design. Furthermore, conduction of such analysis takes the great computational cost. Presented work concerns possibility of application of regression trees, especially CART model, in preliminary analysis for sensitivity analysis. Use of data mining algorithms enables acquiring of preliminary, rough results: relationships among parameters of the hot rolling process of dual phase steel strips and rules of optimization of this process, it also does not require any apriori knowledge about thermomechanical processes.