Health Indicator Construction for Milling Tool Wear Monitoring With Multi-sensor Fusion

Tool wear is an important limitation to milling productivity and machining part quality. This paper presents a new approach for tool condition monitoring (TCM). In the proposed TCM method, multiple domain features are firstly extracted by adaptive decomposition of original multi-sensor monitoring signals. Then these features are evaluated and optimally selected to adaptively integrate into a comprehensive health indicator (HI) using principal component analysis (PCA) for characterization of the wear state of the milling tool. High speed milling data sets from 2010 prognostics and health management (PHM) challenge is studied to verify effectiveness of the proposed TCM approach. The experimental results demonstrate that the proposed method can effectively fuse multi-sensor information to reliably track the online wearing process of milling tool.