This thesis addresses the correlation of the mechanical properties of materials with the value of the machinability number when cutting with an abrasive water jet (AWJ) technology. This connection was researched by machine learning approach. In the theoretical part, the AWJ process is presented. The methods of generating the AWJ, abrasives, as well as the physical principle of removing material and the empirical model for cutting with the AWJ are described. Some mechanical properties of materials that affect the value of the machinability number are presented in more detail. In the main part, machine learning and the use of regression trees in the Matlab software tool are presented. In the practical part, the results are shown through graphs that display the deviation of the regression tree forecast model from the true value. With the machine learning approach it was found that the tensile strength is the most influential input parameter in connection with the machinability number forecast, which is followed by the elastic modulus parameter.
|
