Chatter can limit productivity in milling. The stability lobe diagram, which separates stable axial depth of cut-spindle speed pairs from unstable (or chatter) pairs, can be used to select optimal parameters. However, calculating the stability lobe diagram requires the tool point frequency response function and cutting force model, which may not be available in production environments. In this presentation, a Bayesian machine learning approach for milling stability boundary and optimal parameter identification is described. For this approach, knowledge of the tool frequency response function and force model is not required. Instead, the probability of stability for each axial depth-spindle speed pair is updated with each new test result (stable or unstable). An adaptive experimental strategy to identify operating parameters that maximize material removal rate is described. Results show that the proposed method is an efficient method to identify the stability lobe diagram and optimal operating parameters in a production environment from a limited number of tests.
- Understand the basics of milling stability prediction using a physics-based approach
- Understand the basics of Bayesian machine learning
- Understand how Bayesian machine learning can be applied to select milling parameters for maximum material removal rate
Professor/ORNL Joint FacultyUniversity of Tennessee, Knoxville