Multisensory Chatter Detection in Band Sawing

With the development and application of expensive, difficult to cut metals and metal alloys, the minimization of waste material for final operations has, together with the quality of the band sawing process, become more and more important. As the onset of chatter can have a very detrimental effect on the quality of the cut, on the quality of the resulting surface, and on process performance in general, the prompt detection of chatter is of high importance. In the paper a multisensory approach is investigated for chatter detection in the band sawing process. In the experiments steel workpieces of geometrically different profiles were used. Based on an analysis of the acquired signals of the cutting forces, machine vibrations and emitted sound, a method involving a set of features for the detection of chatter in a cutting regime has been defined. The proposed method is not affected by the workpiece geometry or the process parameters. Analysis of the individual features extracted from the various process signals has been performed for chatter and chatter-free band sawing regime classification. The paper presents the results obtained using a cross-validation approach, and summarizes the most informative extracted features with respect to the various process signals.