Detection of Roughing Tool Defects in Broaching of Fir-Tree Slots for Aerospace Engines Using Process Monitoring Systems

Fir-tree slots provide the form-fit connection between turbine disks and blades in aerospace engines and therefore constitute highly stressed geometries in safety-critical components. These slots are manufactured by broaching, a machining process performed near the end of the manufacturing chain for turbine disks made from high-strength materials such as Inconel 718. The severe thermo-mechanical loads encountered during broaching result in pronounced wear of high-speed steel tools, potentially leading to cutting-edge chipping and, consequently, compromising component quality. Early and reliable detection of tool defects is therefore essential to ensure process stability and component quality. In this study, a data-driven process monitoring system based on internal machine signals was used to investigate the detectability of cutting-edge chipping of broaching tools with straight cutting edges, identifying relevant signal sources and features and evaluating different modeling approaches. Broaching experiments with intentionally induced defects were conducted, and features in the time and frequency domains were extracted from torque, current, and three-phase motor signals. These features were used to train and compare a support vector machine, a random forest, and a neural network. The results demonstrate that reliable classification is feasible, with the neural network and support vector machine achieving balanced accuracies of up to 98.96% and 99.13%, respectively. High-frequency motor signals showed the highest relevance, highlighting their potential for industrial applications.