Feature-Based Analysis of Acoustic Emission Signals for Wear Monitoring in Centerless Through-Feed Grinding

Wear monitoring is a critical aspect of maintaining the health and performance of machinery. During centerless grinding, grinding and regulating wheels as well as work rest blades wear out over time due to contact with workpieces, which impairs the geometrical accuracy and surface quality of the processed parts, and ultimately the productivity of the process. Acoustic emission sensors are a promising source of information for wear monitoring which in turn allows to optimize dressing intervals and prevent rejects. During an experimental series, acoustic emission signals were collected from a centerless through-feed grinding process to identify changes in the signal that could be indicative of wear. The collected acoustic emission signals were preprocessed by digital filtering and extracting a comprehensive set of features. A one-class support vector machine was used to quantify the signal evolution over the span of the experimental series. The resulting resemblance of the signal evolution to the grinding wheel wear effects observed in the geometric properties of the workpieces from the series suggests the validity of this approach.