MACHINE LEARNING FOR STABILITY ANALYSIS OF MILLING PROCESS

The properties related to chatter and the stability of manufacturing milling machines are dependent primarily on parameters such as axial depth of cut, rotary speed rpm, and cutting force. These parameters are used to calculate the milling process and avoid certain cuts that would result in unstable milling. Using applied data science tools for statistical analysis and machine learning models like SHapley Additive exPlanations (SHAP), this research will produce valuable insight into two stability testing methods used in milling machining operations. By leveraging the data, a partitioning model can be used with known stable and unstable points under certain conditions to sort and select the best point(s) to be used in testing on the stability lobe diagram. The aim is to compare the two testing methods of combination and sequencing testing for cost-effectiveness and accuracy in identifying the milling machine’s stability boundary which in turn will help with reducing the amount of chatter present in the milling. Recognizing the testing method with the better accuracy will help determine the best sequence of testing and the stopping criteria.