Improved random forest for titanium alloy milling force prediction based on finite element-driven

Titanium alloys are widely used in various industries, such as aerospace and 3C, because of their high specific strength and heat resistance, leading to the extensive application of specialized cutting tools. The cutting force is a critical factor reflecting the machining state, and the prediction of milling forces can greatly assist in the design and development of cutting tools. Therefore, this paper proposes a prediction model driven by finite element data via the chimp optimization algorithm-random forest (CHOA-RF). The model is trained with tool geometric parameters (unequal helix angle, rake angle, and shape factor) and milling parameters (spindle rotation speed, feed rate, and axial depth of cut) to obtain predicted milling force values. Additionally, cutting tools with different structures were prepared, and milling experiments were conducted. The real experimental data were used as the test set to evaluate the model. Finally, the CHOA-RF model was compared with support vector regression (SVR), the genetic algorithm-optimized back propagation neural network (GA-BP), and the particle swarm algorithm-optimized back propagation neural network (PSO-BP). The results showed that for the prediction of F_y, the CHOA-RF model outperformed SVR, GA-BP, and PSO-BP, with the lowest root mean square error (RMSE) of 2.13, the highest coefficient of determination (R²) of 0.93, and the highest average accuracy (ACC) of 96.23% on the validation set.