A Random Forest Regression Model for Predicting Residual Stresses and Cutting Forces Introduced by Turning IN718 Alloy

Random Forest algorithm has been utilized in this paper for creating a regression model to predict residual stresses and cutting forces introduced by turning IN718 alloy. Residual stresses and cutting forces are crucial to deformations of thin walled parts. Meanwhile, it is difficult to determine a specific regression equation expression between the dependent variables (surface residual stresses, max compressive residual stresses, cutting forces) and independent variables (depth of cut, feed rate, cutting speed). Random forest algorithm has the superiority as it can build regression relation between variables using many regression trees rather than a specific equation. Multiple sets of simulations were finished by the finite element method to obtain the regression training data. A series of turning and residual stresses measurement experiments with different turning parameters were conducted to ensure the accuracy of the regression training data. From the research, predicted residual stresses and cutting forces were consistent with simulation results, and the prediction errors were controlled within the acceptable range. The research could contribute the further investigation of thin-walled part deformation control.