Chatter stability boundary analysis using RBNN

Abstract

An analytic realization of a chatter mechanism using an experimental force data with the radial basis neural network (RBNN) was introduced and compared with a conventional stability analysis. In this regard, the FFT and time series spectrum analysis was used for defining the criteria as the existence of chatter with the end milling force and the desired coded output of chatter was trained and finally converged to desired outputs with RBNN. The output of the stability lobe with the RBNN matches well with the conventional desired stability lobe. Using this trained learning algorithm, the stability boundary with RBNN was acquired using the contour plotting. As a result, the proposed stability lobe boundary with the neural network also consists with the conventional stability boundary lobe that is calculated in the characteristic equation of a transfer function in the chatter dynamics. In the RBNN stability analysis, two input and three output parameters were used for configuring the input and output vectors in neural network.