Hybrid Ultrasonic-Assisted Milling with Internal Cooling and Ensemble Machine Learning for Cutting-Tool Wear Prediction in SKD11 Hardened Steel

Machining of SKD11 tool steel challenging task due to high hardness and strength of material, which are responsible for accelerated cutting tool wear, high cutting temperature, and poor surface finish. The traditional cooling system is not able to control the heat generation process at higher speeds in an effective manner. Also, the previous research work related to Ultrasonic-Assisted milling has focused on restricted scope. This study work is concentrated on the development of an improved cutting strategy which enhances the life of the cutting tool and the quality of the surface. This work approach combines Ultrasonic-Assisted milling with through-tool internal cooling method. The ultrasonic vibration generates a high-frequency, low-amplitude motion which reduces the friction and improves chip removal. However, the internal cooling performs coolant injection directly into the cutting zone for better heat dissipation. Comparative experimental tests between Z-axis, XY-axis, and XYZ-axis ultrasonic mode assisted milling under different cooling conditions proved that the best results were obtained by Z-axis ultrasonic assistance and internal cooling: surface roughness was decreased by 20–31%, and the temperature by 15–28%, as compared to conventional milling. Various machine learning (ML) models, such as Decision Tree, Random Forest, and AdaBoost, are used to predict cutting tool wear. Among these, AdaBoost provides the highest accuracy and precision values i.e., 0.94 and 0.964286 respectively, which is more improved by a simple ensemble vote. This integrated process and predictive model offer a practical solution to improve the machinability of SKD11 and enable smarter, more sustainable manufacturing.