Identification of Milling Stability by using Signal Analysis and Machine Learning Techniques

Abstract

Nowadays, milling stability is one of the most concerns in the manufacturing industry in order to reduce the cost of tool replacement, increase the productivity as well as increase precision and surface quality of the metal cutting process. One of the most critical components of the machining process is to identify chatter during the cutting process. This paper proposed a cutting signal processing methodology to create, analyze, and select relevant features for the chatter identification. An effective technique based on feature learning was proposed to monitor the cutting stability using vibration signals. The technique of signal transformation such as Fourier Transform (FT) is an effective tool to determine the frequencies related to machining operation and cutting stability. Machine learning models such as Random Forest, Decision Tree, and eXtreme Gradient Boosting were applied for selected features as the step of classification. The success of this proposed method was proved by solid statistical support for the features selection method and the performance of Random Forest achieved 98% of accuracy at two states of milling process including stable, and unstable conditions.