Abstract
Chatter is a common phenomenon in micromachining processes that adversely affects machining quality, reduces tool life, and generates excessive noise that contributes to environmental pollution. Therefore, the timely detection of chatter is crucial for sustainable production. This paper presents an investigation on the extraction of two types of features, i.e., probability-related and entropy-related, using Shannon entropy and Rényi entropy algorithms, respectively, for chatter detection in micro milling. First, four chatter features were examined using actual machining tests under stable, weak-chatter, and severe-chatter conditions. Second, the proposed chatter features were systematically assessed by combining the characteristic change rates, threshold intervals, and computation times. The results demonstrated that the proposed features can effectively detect the occurrence of chatters at various severity levels. It was found that the probability-related features exhibit better sensitivity compared to entropy-related features, and the features extracted from Shannon entropy algorithm are more sensitive than the Rényi entropy algorithm.