Chatter Detection in Milling Machines by Neural Networks classification and feature selection

In modern industry, milling is an important tool when a high material removal rate is required. Chatter detection in this situation is a crucial step for improving surface quality and reducing both noise and rapid wear of the cutting tool. This paper proposes a new methodology for the chatter detection in CNC milling machines. This methodology is based on vibratory signal analysis and artificial intelligence. The methodology consists in five major steps: 1) data acquisition, 2) signal processing, 3) features generation, 4) features selection and 5) classification. As chatter components occur around system resonance frequencies, a multiband resonance filtering method is proposed at the processing step. The process is then followed by envelope analysis. This allows for increasing the signal-to-noise ratio and increasing the sensitivity of generated features. Extracted features are then ranked based on their entropy in which only best features are selected and presented to the system for classification. At the classification step, the selected features are classified into two classes: stable and unstable utilizing neural networks. Two neural network approaches, Radial Basis Function (RBF) and Multi-Layer Perceptrons (MLP), are tested. The developed approach is applied for chatter detection in Huron K2X10 milling machine. This approach is tested on a milling machine at different depths of cut and various rotational speeds. Discussions are made and the results confirm the accuracy of the proposed methodology.