Identification of machining state in robot milling based on multi-characteristics gray correlation analysis

Abstract

Mode coupling chatter and regenerative chatter are the two important types of robot milling chatter. The accurate identification of machining state of robot milling is essential for suppressing milling chatter and improving machining quality. In this paper, the robot milling state identification model based on multi-characteristics gray correlation analysis is proposed. First, the frequency removal algorithm (FRA) is used to process the original signal. Then, based on the improvement of traditional correlation degree, combined with the difference level classification, this paper presents a multi-characteristics gray correlation analysis. This paper uses this method to select the highly correlated chatter characteristic set for stable milling and chatter type differentiation from the initial chatter characteristic set. Then, improved principal component analysis (IPCA) is used to analyze the highly correlated chatter characteristic set and obtain the multivariate fusion chatter characteristic. Subsequently, the Gaussian Mixture Model (GMM) was employed to solve for the identification thresholds, resulting in the fusion feature threshold for distinguishing between stable milling being −845.3 and the fusion feature threshold for differentiating between types of chatter being 972.8. Finally, it is verified by the sample data that the model can accurately identify stable milling, mode coupling chatter and regenerative chatter, and it also has strong anti-noise capabilities. This provides theoretical guidance for the targeted suppression strategy of robot milling chatter.