Novel sensor-based tool wear monitoring approach for seamless implementation in high speed milling applications

Intelligent Cyber-Physical Monitoring and Control of I4.0 Machining Systems - An Overview and Future Perspectives

“…This approach provides a cutting conditions-independent representation of tool condition, and thus, significantly reduces the required learning effort. An N-way ANOVA analysis of the WSCNN extracted features has shown the features' high sensitivity to the tool condition compared to the cutting conditions (feedrate and depth of cut), tool geometry (number of flutes, diameter, and corner radius) and their interactions [13]. The features provided a distinguished and stable representation of tool wear state over time.…”

“…This limits the capability of the features extracted directly from raw signals to globally represent the tool wear state. Recently, Hassan et al [13] developed a novel signal processing and decision making approach that can mask the effect of cutting parameters and accentuate the tool wear effect in the extracted features. Additionally, this approach greatly minimizes the learning effort needed to train the decision making model and reduces the signal noises effect on the extracted features, which enables its implementation in industrial environment.…”

“…In [13], the developed approach has been applied to signals acquired from a high-end triaxial accelerometer mounted on the spindle head near the cutting tool, which was wired to a National Instrument data acquisition card type NI 9232. It discussed the system applicability to conditions far from the training data and demonstrated the approach capability to accurately detect the tool condition at different cutting conditions.…”

“…A BECKHOFF CX8190 industrial computer with an 800 MHz ARM Cortex™-A9 processor was used to acquire and store the spindle feedback signals from the EL3783 card through EtherCAT connection. Figure 5 shows a time-frequency scattergram of the WSCNN features extracted using the approach developed by the authors in [13] from fresh and worn tools using three different sensors namely, a PCB accelerometer (model 356A71), the developed self-powered wireless cost-effective accelerometer, and the BECKHOFF current transducer. The shown scattergram was calculated for tool T1 during a side-milling test at fz = 0.1 mm/tooth, ap = 7 mm, and ae = 12 mm.…”

“…Acoustic emission AE sensors can capture abnormalities during the turning or milling operations, and have the flexibility to be mounted on the tool or the workpiece without major disturbances to various machining activities [11,12]. Worn and broken tools can cause high vibrations during cutting operations; these vibrations can be detected using accelerometers [13][14][15] Additionally, temperature and ultrasonic sensors have been used to monitor the tool condition and surface roughness, respectively [16][17][18]. Nevertheless, they did not show their readiness to be applied in industrial applications.…”