講演情報
[14p-K508-2]Wavelet-LSTM Approach for Optical Surface Extension
〇(D)Ke Chen1, Hayasaki Yoshio1 (1.Utsunomiya Univ. for Utsunomiya University)
キーワード：
polishing、neural network、material removal
This study aims to address the edge effect problem commonly encountered in contact machining processes. By employing wavelet multiscale decomposition, the surface profile's frequency information is separated, and the separated frequency components are predicted using an LSTM neural network. The extended frequency components are then utilized to achieve surface profile extrapolation through reconstruction. Compared with traditional extrapolation methods, the proposed approach significantly improves the quality of the machined surface, offering a novel solution for high-precision machining.