ABSTRACT
This work presents a method that deals with the problem of determining tool wear, surface roughness, and cutter temperate in the machining of a carbon fibre-reinforced polymer (CFRP) workpiece without the need for removing the tool for inspection. The spindle acceleration signature is dominated by the cutter passing frequency and its first harmonic. Cutting parameters and tool conditions strongly correlate with shifts in the latter two frequencies. The kriging method is utilised to construct prediction models of tool conditions using cutting parameters and spectral shifts. The presented Kriging-based method is unique in its ability to generate acceptable predictions of tool conditions using a small set of experimental data to construct the prediction models. Validation experiments have shown that the proposed method predicted tool wear, surface roughness, and cutter temperature with accuracies of 91%, 86%, and 84%, respectively. Benchmarking of the proposed method against another prediction method has shown acceptable performance.
Highlights
This work utilises spectral shifts of the spindle acceleration to predict tool condition and surface quality of the workpiece
Kriging models are constructed based on a small experimental data set. Thus, this method can be used most effectively when experimental data is scarce,
Benchmarking the proposed method has shown acceptable performance in comparison with well-known methods such as ANN
Proposed method is adaptable to the real-time prediction of tool condition and surface quality
Acknowledgments
Support of the American University of Iraq – Sulaimani and Khalifa University is acknowledged
Disclosure statement
No potential conflict of interest was reported by the author(s).