A prior-knowledge input LSSVR metamodeling method with tuning based on cellular particle swarm optimization for engineering design

作者：

Highlights：

• A prior-knowledge input LSSVR (PKI-LSSVR) is proposed.

• In PKI-LSSVR, the low-fidelity knowledge is used as the input of LSSVR.

• CPSO is adopted with a new neighborhood function to tune the hyper-parameters.

• Case study shows PKI-LSSVR performs well in accuracy and robustness.

摘要

•A prior-knowledge input LSSVR (PKI-LSSVR) is proposed.•In PKI-LSSVR, the low-fidelity knowledge is used as the input of LSSVR.•CPSO is adopted with a new neighborhood function to tune the hyper-parameters.•Case study shows PKI-LSSVR performs well in accuracy and robustness.

论文关键词：Variable fidelity metamodel,Support vector regression,Cellular particle swarm optimization,Design optimization

论文评审过程：Available online 17 September 2013.

论文官网地址：https://doi.org/10.1016/j.eswa.2013.09.010