Sensor Signal Segmentation for Tool Condition Monitoring

Abstract Robust tool condition monitoring system requires reliable, repeatable selection of representative segments of the sensor signals. In commercial TCM systems and most of laboratory ones useful signal segments are selected by the system user, which is difficult, inconvenient, prone to random changes of cutting conditions and human errors. The paper presents algorithms for fully automatic detection of actual cutting (elimination of air cutting), selection of relatively stable signal segments representative of the tool condition and elimination of the overabundance of signal data in case of long operations or tool lives.