Tool-holder integrated printed piezoceramic sensors for process state classification and tool-wear progress evaluation in turning

Abstract

Tool condition monitoring strategies in turning processes, using multiple sensor sources to measure dynamic quantities (e.g., force, acoustic emissions, acceleration), has gained significant attention in conjunction with machine learning methods. As a novel approach, the introduction of printed piezoceramic thick films in machine tools allows the customization of sensor shape and placement, optimized to fit the tool holder’s geometry and process boundary conditions. This work proposes integration concepts using two configurations of printed piezoceramic sensors. This includes the use of a finite element simulation to estimate the generated electric charge from sensors on three locations on the tool holder, induced by transient mechanical loading. The design is validated under controlled mechanical loading and tested across various cutting conditions and tool-wear states. Signal information content from each sensor location, relative to different machining conditions, is assessed through recursive feature elimination and validated using dimensionality reduction. Finally, tool-wear progress is evaluated using the selected features.