Parallel method speeds up multi-fidelity engineering optimization
Parallel Multi-Fidelity Expected Improvement Method for Efficient Global Optimization
Computational Engineering, Finance, and Science
Summary
When designing complex things like turbine blades, engineers use computer models that vary in accuracy and cost. High-accuracy models are expensive to run, while simpler ones are cheap but less precise. The authors propose a new way to smartly decide when to use each kind of model during optimization, balancing thorough searching with saving time. Their method also runs multiple simulations in parallel, making the process faster. Tests show it works well on both simple math problems and real engineering design.
What this means in practice
- •For engineering design teams: Optimize designs faster by jointly using detailed and simplified simulations with efficient sample allocation.
- •For manufacturing process planners: Reduce computational costs in complex process simulations by adaptively balancing accuracy and computation time.
Authors
Zhendong Guo, Qineng Wang, Liming Song, Jun Li
Abstract
Multi-Fidelity optimization (MFO) has received extensive attentions in engineering design, which resorts to augmenting the small number of expensive high-fidelity (HF) samples by a large number of low-fidelity (LF) but cheap samples to improve the optimization performance. A key factor that influences the effectiveness of MFO is how to adaptively assign samples for HF and LF simulations in the iteration process. To address such sample assignment issue in MFO, we propose a new infill criterion named as Filter-GEI, which imposes an adaptive filter function on top of the generalized expected improvement (GEI) acquisition function. In particular, by taking the correlations between HF and LF models into account, the Filter-GEI can efficiently allocate HF and LF samples to achieve a good balance in between the local and global search. Furthermore, considering parallel computing, the Filter-GEI infills multiple HF and LF samples in each iteration, which can further improve its efficiency as computing power increases. Through tests on five mathematical toy problems and one engineering problem for the turbine blade design, the effectiveness of the proposed algorithm has been well demonstrated.