Bayesian Inference for Milling Stability Modeling

Abstract

This essay describes Bayesian learning for milling stability modeling. A non-model grid-based method and two model-based methods using random sample stability maps and Markov Chain Monte Carlo sampling for Bayesian learning are described. The three methods are compared using experimental results completed on Aluminum 6061-T6 workpiece. A test selection strategy to maximize the material removal rate is presented for the non-model and model-based approaches. The essay also describes recent advances in Bayesian learning and experimentations and provides future research directions and outlook.