TY - GEN
T1 - A New Prediction Method of Tool Life Considering Cognitive Uncertainty by Delayed Rejection Adaptive Sampling
AU - Wang, Zenghui
AU - Zhou, Guanghui
AU - Zhang, Chao
AU - Chang, Fengtian
AU - Zhou, Yaguang
AU - Zhao, Dan
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Aiming at the problem of cognitive uncertainty caused by lack of knowledge reserve, imperfect data information, and subjective experience in the traditional tool life prediction, a new tool life prediction method based on Bayesian inference is proposed. Firstly, considering the inherent uncertainty of the empirical constants in extension Taylor tool life equation, the prior distribution of the uncertainty parameters is determined according to expert experience and historical data in knowledge base. Then, the delayed rejection and adaptive sampling strategy is introduced to improve the acceptance rate of candidate samples, and enable the proposal distribution to perform adaptive sampling in Bayesian inference. Based on the Taylor equation updated by Bayesian inference, the posterior distribution of tool life is estimated by Monte Carlo simulation and statistical analysis. Finally, the proposed method is verified by turning tests of cemented carbide tools and MS309 steel. The test results show that the obtained Markov chain has better overall performance and higher sample acceptance rate. The tool life prediction can better approximate the experimental value of each group, which can effectively reduce the influence of cognitive uncertainty on tool life prediction.
AB - Aiming at the problem of cognitive uncertainty caused by lack of knowledge reserve, imperfect data information, and subjective experience in the traditional tool life prediction, a new tool life prediction method based on Bayesian inference is proposed. Firstly, considering the inherent uncertainty of the empirical constants in extension Taylor tool life equation, the prior distribution of the uncertainty parameters is determined according to expert experience and historical data in knowledge base. Then, the delayed rejection and adaptive sampling strategy is introduced to improve the acceptance rate of candidate samples, and enable the proposal distribution to perform adaptive sampling in Bayesian inference. Based on the Taylor equation updated by Bayesian inference, the posterior distribution of tool life is estimated by Monte Carlo simulation and statistical analysis. Finally, the proposed method is verified by turning tests of cemented carbide tools and MS309 steel. The test results show that the obtained Markov chain has better overall performance and higher sample acceptance rate. The tool life prediction can better approximate the experimental value of each group, which can effectively reduce the influence of cognitive uncertainty on tool life prediction.
KW - Bayesian inference
KW - adaptive sampling
KW - cognitive uncertainty
KW - delayed rejection
KW - tool life prediction
UR - https://www.scopus.com/pages/publications/85174416951
U2 - 10.1109/CASE56687.2023.10260669
DO - 10.1109/CASE56687.2023.10260669
M3 - 会议稿件
AN - SCOPUS:85174416951
T3 - IEEE International Conference on Automation Science and Engineering
BT - 2023 IEEE 19th International Conference on Automation Science and Engineering, CASE 2023
PB - IEEE Computer Society
T2 - 19th IEEE International Conference on Automation Science and Engineering, CASE 2023
Y2 - 26 August 2023 through 30 August 2023
ER -