Robust stability prediction in milling process considering the correlations of input parameters

Abstract

Because of the uncertainties of the input parameters in dynamic milling model, robust stability prediction is a technology to effectively avoid the occurrence of regenerative chatter. However, the existing robust stability prediction methods generally consider the potential interval of each input parameter as their uncertainty measurements directly, without taking into account the correlations between input parameters. In this regard, a novel robust stability prediction method considering the correlations between input parameters is proposed in this paper. To consider the correlations between input parameters, high-dimensional random variables composed of input parameters are constructed, and their probability density functions are estimated according to the observation data. Based on the obtained probability density functions, the confidence regions of these random variables are determined and used to describe the uncertainty boundaries of input parameters. The problem of judging the robust stability of milling at a given speed and axial depth is transformed into an optimization problem, and the confidence regions are used to construct the constraints for the optimization variable. The experimental results have shown that the robust stability boundary generated by the proposed method can effectively avoid the occurrence of regenerative chatter and ensure the stability of milling operation.