Process model-based approach for machine tool and cutting process diagnostics

The desire to improve product quality while at the same time improving process productivity has led to manufacturing process strategies that are much more diagnostic in their nature. This paper focuses on two aspects of this dilemma for machining process: (i) the modeling of common process faults that will enable the process model-based approach to both off-line and on-line data analysis and fault diagnosis and, (ii) fault recognition in process cutting force data for both singly-occurring and multiple faults. The paper defines several process faults included in an end milling model. These include: flute breakage, chipping, parallel-axis offset runout, spindle tilt, and individual flute run-out/deviation. An experiment has been performed that examines the nature of faults in the cutting force signal as a function of tool geometry, cutting conditions, and cut geometry parameters and includes the study of several parametric measures/indices for process fault diagnosis. To facilitate the fault diagnosis process, analysis techniques in the time domain and frequency domain are employed. The results show that the measures/indices proposed herein do provide discriminatory power to uniquely identify process faults.