Probabilistic Sequential Prediction of Cutting Force Using Kienzle Model in Orthogonal Turning Process

Abstract

Probabilistic sequential prediction of cutting forces is performed applying Bayesian inference to Kienzle force model. The model uncertainties are quantified using the Metropolis algorithm of the Markov chain Monte Carlo (MCMC) approach. Prior probabilities are established and posteriors of the models parameters and force predictions are completed using the results of orthogonal turning experiments. Two types of tools with chamfer (rake) angles of 0 deg and −10 deg are tested under various cutting speed and feed per revolution values. First, Bayesian inference is applied to two force models, Merchant and Kienzle, to investigate the cutting force prediction at the low feed values for the 0 deg rake angle tool. Second, the results of the posteriors of the Kienzle model parameters are used as prior probabilities of the −10 deg rake angle tool. The simulation results of the 0 deg and −10 deg tool rake angle are compared with the experiments which are obtained under other cutting conditions for model verification. Maximum prediction errors of 7% and 9% are reported for the tangential and feed forces, respectively. This indicates a good capability of the Bayesian inference for model parameter identification and cutting force prediction considering the inherent uncertainty and minimum input experimental data.