TY - GEN
T1 - Cutting Force Similarity Calculation in Milling Process Using Siamese LSTM Structure
AU - Kwak, Juheon
AU - Jo, Wonkeun
AU - Lee, Soomin
AU - Kim, Hyein
AU - Koo, Jeongin
AU - Kim, Dongil
N1 - Funding Information:
ACKNOWLEDGMENT This study was supported by the Korea Institute of Industrial Technology’s learning-based process ability diagnostic control system development project (Kitech EO-19-0043) for self-optimization of production systems. In addition, it received support from the Information and Communication Planning and Evaluation Institute (IITP-2019-0-01343 (Graduate School of Convergence Security)) and the National Research Foundation (No. 2020R1F1A1075781) with the finances of the Ministry of Science and ICT.
Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Cutting force is a key factor in machining processes. Cutting force similarity is required for several important issues, namely: stability evaluation, process control, and process parameter setting. This study employed a long short-term memory (LSTM) with Siamese architecture to measure the similarity of the cutting forces in a milling process. The Siamese LSTM was trained with time series data of the vertical cutting force collected from a cutting tool during the milling process to calculate the similarity. For evaluation, dynamic time warping (DTW), a common approach used to calculate the similarity of time series data, was employed for comparison with the Siamese LSTM. Experimental results showed that the proposed Siamese LSTM outperformed the conventional DTW-based similarity calculation.
AB - Cutting force is a key factor in machining processes. Cutting force similarity is required for several important issues, namely: stability evaluation, process control, and process parameter setting. This study employed a long short-term memory (LSTM) with Siamese architecture to measure the similarity of the cutting forces in a milling process. The Siamese LSTM was trained with time series data of the vertical cutting force collected from a cutting tool during the milling process to calculate the similarity. For evaluation, dynamic time warping (DTW), a common approach used to calculate the similarity of time series data, was employed for comparison with the Siamese LSTM. Experimental results showed that the proposed Siamese LSTM outperformed the conventional DTW-based similarity calculation.
KW - cutting force
KW - dynamic time warping
KW - long short-term memory
KW - Manhattan distance
KW - Siamese neural network
KW - similarity
UR - http://www.scopus.com/inward/record.url?scp=85161596480&partnerID=8YFLogxK
U2 - 10.1109/MEEE57080.2023.10126810
DO - 10.1109/MEEE57080.2023.10126810
M3 - Conference contribution
AN - SCOPUS:85161596480
T3 - 2023 2nd International Conference on Mechatronics and Electrical Engineering, MEEE 2023
SP - 54
EP - 58
BT - 2023 2nd International Conference on Mechatronics and Electrical Engineering, MEEE 2023
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2nd International Conference on Mechatronics and Electrical Engineering, MEEE 2023
Y2 - 10 February 2023 through 12 February 2023
ER -