A novel approach of shot peening process parameters prediction with missing surface integrity data based on imputation method

The process parameters of shot peening can be actively predicted by a large number of surface integrity data after treatments. However, surface integrity data are not complete in engineering practice. This study proposes a novel approach to shot peening process parameters prediction with missing sur...

Full description

Saved in:
Bibliographic Details
Published in:International journal of advanced manufacturing technology 2023-07, Vol.127 (1-2), p.81-92
Main Authors: Li, Yang, Wei, Peitang, Zhao, Xinhao, Zhu, Rupeng, Wu, Jizhan, Liu, Huaiju
Format: Article
Language:English
Subjects:
Accuracy
CAE) and Design
Computer-Aided Engineering (CAD
Engineering
Industrial and Production Engineering
Integrity
Mathematical models
Mechanical Engineering
Media Management
Microhardness
Missing data
Original Article
Performance evaluation
Process parameters
Residual stress
Shot peening
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The process parameters of shot peening can be actively predicted by a large number of surface integrity data after treatments. However, surface integrity data are not complete in engineering practice. This study proposes a novel approach to shot peening process parameters prediction with missing surface integrity data. The imputation method adopted multidimensional information of missing surface integrity data and is utilised to improve the accuracy and robustness of the proposed model. The results indicate that under the missing rate of residual stress and micro-hardness of 20%, the proposed approach for shot peening intensity, coverage, and shot diameter has significantly improved prediction accuracy compared to a model without missing data imputation, with rises of 8.63%, 24.99% and 30.92%, respectively. Additionally, the influence of different missing data proportions of residual stress and micro-hardness on prediction accuracy was investigated. It is revealed that when missing proportions of the residual stress and micro-hardness remain the same, the maximum prediction error occurs, although this error is relatively low when the proportion of missing micro-hardness data is high. The proposed approach has good versatility and can be extended to parameter prediction and performance evaluation of various processes.
ISSN:0268-3768
1433-3015
DOI:10.1007/s00170-023-11514-x