TY - JOUR
T1 - An improved case based reasoning method and its application in estimation of surface quality toward intelligent machining
AU - Xu, Longhua
AU - Huang, Chuanzhen
AU - Li, Chengwu
AU - Wang, Jun
AU - Liu, Hanlian
AU - Wang, Xiaodan
N1 - Funding Information:
This work is financially supported by National Natural Science Foundation of China (51675312, 51675313).
Publisher Copyright:
© 2020, Springer Science+Business Media, LLC, part of Springer Nature.
PY - 2021/1
Y1 - 2021/1
N2 - In the high speed milling process, the accurate predictions of surface roughness and residual stress can avoid the deterioration of machined surface quality. But it’s hard to estimate the surface roughness and residual stress under different tool wear status and cutting parameters. In this work, a novel intelligent reasoning method-improved case based reasoning (ICBR) was proposed to predict the surface roughness and residual stress. The inputs of ICBR are cutting parameters and tool wear status. The corresponding outputs of ICBR are surface roughness and residual stress. In the ICBR, K-nearest neighbor method and artificial neural network (ANN) as case retrieval was introduced to retrieve the K similar cases to the inputs. Through retrieving K similar cases, the Gaussian process regression (GPR) model as case reuse was established to output the surface roughness and residual stress. The vibration particle swarm optimization algorithm is proposed to optimize the ANN and GPR models. The high speed milling experiments of Compacted Graphite Iron was performed to validate the performance of ICBR. The experimental results showed that the cutting speed is the most important factor affecting the surface roughness. The feed rate is the most important factor affecting the residual stress. The ICBR gives the accurate estimation of surface roughness with the Mean Absolute Percentage Error of 11.6%. As for residual stress, the prediction accuracy using ICBR is 87.5%. Compared with Back-Propagation neural network, standard CBR and GPR models, the ICBR has better predictive performance and can be used for estimations of surface roughness and residual stress in the actual machining process.
AB - In the high speed milling process, the accurate predictions of surface roughness and residual stress can avoid the deterioration of machined surface quality. But it’s hard to estimate the surface roughness and residual stress under different tool wear status and cutting parameters. In this work, a novel intelligent reasoning method-improved case based reasoning (ICBR) was proposed to predict the surface roughness and residual stress. The inputs of ICBR are cutting parameters and tool wear status. The corresponding outputs of ICBR are surface roughness and residual stress. In the ICBR, K-nearest neighbor method and artificial neural network (ANN) as case retrieval was introduced to retrieve the K similar cases to the inputs. Through retrieving K similar cases, the Gaussian process regression (GPR) model as case reuse was established to output the surface roughness and residual stress. The vibration particle swarm optimization algorithm is proposed to optimize the ANN and GPR models. The high speed milling experiments of Compacted Graphite Iron was performed to validate the performance of ICBR. The experimental results showed that the cutting speed is the most important factor affecting the surface roughness. The feed rate is the most important factor affecting the residual stress. The ICBR gives the accurate estimation of surface roughness with the Mean Absolute Percentage Error of 11.6%. As for residual stress, the prediction accuracy using ICBR is 87.5%. Compared with Back-Propagation neural network, standard CBR and GPR models, the ICBR has better predictive performance and can be used for estimations of surface roughness and residual stress in the actual machining process.
KW - ANN model
KW - GPR model
KW - ICBR method
KW - VPSO algorithm
UR - http://www.scopus.com/inward/record.url?scp=85083589603&partnerID=8YFLogxK
U2 - 10.1007/s10845-020-01573-2
DO - 10.1007/s10845-020-01573-2
M3 - Article
AN - SCOPUS:85083589603
SN - 0956-5515
VL - 32
SP - 313
EP - 327
JO - Journal of Intelligent Manufacturing
JF - Journal of Intelligent Manufacturing
IS - 1
ER -