Statistical Analysis and Predictive Modelling of Chip Temperature in the Cutting Zone for Tools with Different Rake Angles
Artykuł w czasopiśmie
MNiSW
0
brak dyscyplin
| Status: | |
| Autorzy: | Zagórski Ireneusz, Kulisz Monika, Skoczylas Agnieszka, Zgórniak Piotr, Legutko Stanisław |
| Dyscypliny: | |
| Aby zobaczyć szczegóły należy się zalogować. | |
| Rok wydania: | 2026 |
| Wersja dokumentu: | Drukowana | Elektroniczna |
| Język: | angielski |
| Numer czasopisma: | 21 |
| Wolumen/Tom: | 19 |
| Numer artykułu: | 4728 |
| Strony: | 1 - 20 |
| Impact Factor: | 3,7 |
| Web of Science® Times Cited: | 0 |
| Scopus® Cytowania: | 0 |
| Bazy: | Web of Science | Scopus |
| Efekt badań statutowych | NIE |
| Finansowanie: | This research was funded by Lublin University of Technology within the framework of internal grants for the discipline of Mechanical Engineering FD-20/IM-5/138, FD-20/IM-5/061 and FD-20/IM-5/107. |
| Materiał konferencyjny: | NIE |
| Publikacja OA: | TAK |
| Licencja: | |
| Sposób udostępnienia: | Witryna wydawcy |
| Wersja tekstu: | Ostateczna wersja opublikowana |
| Czas opublikowania: | W momencie opublikowania |
| Data opublikowania w OA: | 6 listopada 2026 |
| Abstrakty: | angielski |
| Experimental thermographic data from dry rough milling of AZ31B and AZ91D magne- sium alloys using tools with rake angles of 5° and 30° were reanalysed using descriptive statistics and predictive modelling. Depending on the assumptions of normality and ho- mogeneity of variance, a one-way ANOVA, Welch ANOVA or Kruskal–Wallis test was applied to up to ten highest available readings of each run; as these are repeated measure- ments within a run, the tests are interpreted descriptively and supplemented with effect sizes. Descriptive comparisons showed differences in the selected upper-temperature readings across parameter levels within the alloy–rake-angle groups. To provide a con- sistently defined upper-tail response, whose estimation precision depends on the number of recorded frames, the 95th percentile (T95) of the thermographic observations was adopted as the predictive response. Gaussian process regression (GPR) was evaluated us- ing leave-one-out cross-validation and a clipped jackknife+-type procedure. The cutting- speed paths had the lowest mean absolute prediction error (9.7 °C), compared with 17.7 °C for feed per tooth and 18.7 °C for axial depth of cut. The nominal 95% GPR prediction intervals achieved only 63.4% mean empirical coverage, increasing to 85.2% with the clipped procedure, which has no standard coverage guarantee and remains below the nominal level. Linear and quadratic regression provided comparable point-prediction ac- curacy and, with the same clipped procedure, comparable coverage (85.2% and 86.6%), whereas a shallow neural network performed worse. Increased model complexity did not consistently improve prediction in this dataset. These results demonstrate that assessment of chip-temperature models requires out-of-sample validation and explicit calibration of predictive uncertainty. |
