A support vector regression-based method for modeling geometric errors in CNC machine tools

For the problem of geometric error prediction of CNC machine tools, an improved hybrid grey wolf optimization (IHGWO) algorithm is proposed to optimize the geometric error modeling scheme of the support vector regression machine (SVR). The predicted and measured values of the geometric error are com...

Full description

Saved in:
Bibliographic Details
Published in:International journal of advanced manufacturing technology 2024-03, Vol.131 (5-6), p.2691-2705
Main Authors: Zhang, Chuanjing, Liu, Huanlao, Zhou, Qunlong, Wang, Yulin
Format: Article
Language:English
Subjects:
Algorithms
CAE) and Design
Computer-Aided Engineering (CAD
Engineering
Error analysis
Error compensation
Error reduction
Euclidean geometry
Industrial and Production Engineering
Machine learning
Machine tools
Mechanical Engineering
Media Management
Modelling
Numerical controls
Optimization
Original Article
Particle swarm optimization
Performance prediction
Screw theory
Support vector machines
Vectors (mathematics)
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:For the problem of geometric error prediction of CNC machine tools, an improved hybrid grey wolf optimization (IHGWO) algorithm is proposed to optimize the geometric error modeling scheme of the support vector regression machine (SVR). The predicted and measured values of the geometric error are combined to construct the fitness function. In IHGWO, principles of particle swarm optimization (PSO) algorithm and dimension learning-based hunting (DLH) search strategies are introduced while retaining the excellent grey wolf position of the basic grey wolf optimization (GWO) algorithm. IHGWO algorithm uses Euclidean distance to construct the neighborhood of individual grey wolves, which enhances the ability to communicate between individual grey wolves and improves the convergence speed and accuracy of the algorithm. Predictive performance of SVR models using sum squared residual to quantify geometric error. Based on the screw theory, space models of geometric errors of CNC machine tools are established and combined with SVR models of geometric errors for compensation. Empirical evidence proves that the proposed method surpasses current error modeling methods in terms of precision and efficiency, as evidenced by a minimum reduction of 9% in circular trajectory error and a reduction to two overruns in S-shaped test pieces after error compensation. This research contributes to the field of CNC machine tool error modeling and has practical implications for manufacturing industries.
ISSN:0268-3768
1433-3015
DOI:10.1007/s00170-023-12212-4