Cutting force prediction by machine learning

This thesis investigates the application of machine learning methods to predict cutting forces by blending them with simulation data using the finite element method (FEM). It analyzes the prediction capabilities and prediction accuracy using various machine learning methods (neural networks, decision trees). The research part emphasizes the importance of machine learning methods in optimizing process parameters, increasing efficiency, and reducing tool wear. In my study, it was shown that the data generated should be processed before learning and transformed as required in terms of the quality of the data generated and the suitability of machine learning methods. In order to provide the most accurate estimates, processes such as removing unnecessary parameters in the data, normalization and correlation analysis were used. As a result of these, it showed the importance of the relationship between the cutting forces and process parameters of the algorithms, that this relationship can be modeled and applied in production areas for process optimization. At the same time, this research emphasizes the importance of integrating cutting force prediction with machine learning into industrial systems. The results show that expanding the dataset and collecting data from production sites will pave the way for progress in issues such as increased efficiency. During the analysis, workpiece material titanium was selected.