Wear estimation in micro-end-milling with wavelet transformations and probabilistic neural networks Article

Tansel, IN, Bao, WY, Arkan, TT et al. (1998). Wear estimation in micro-end-milling with wavelet transformations and probabilistic neural networks . 1998 755-760.



cited authors

  • Tansel, IN; Bao, WY; Arkan, TT; Shisler, B

abstract

  • The original Neural Network-based Periodic Tool Inspection (N2PTI) method performs identical slot milling operations on a test piece with certain time intervals and estimates tool wear from the variation of feed and thrust direction cutting forces. To minimize the influence of depth of cut on estimations and to reduce the training time, use of wavelet transformations and probabilistic neural networks are proposed to estimate usage (wear) of micro-end-mills. Tool usage was estimated with less than 30% error when the system was tested at the usage (wear) levels, which were not used during the training stage. The accuracy of the method is satisfactory to detect the prefailure phase.

publication date

  • December 1, 1998

start page

  • 755

end page

  • 760

volume

  • 1998