Assessment of AVT-331 MF methods
File(s)
Published version
Author(s)
Type
Chapter
Abstract
The multi-fidelity methods studied in AVT-331 are evaluated using a number of assessment categories thought by the team to be important for understanding the general applicability of multi-fidelity methods to vehicle design. These assessment categories are identified, along with the definition of the key mathematical characteristics of AVT-331 benchmark objective functions. Assessment is performed separately for L1, air, sea, and space benchmarks using a common evaluation process. While the process is quantitative, scoring is based on subject matter expertise and consequently subjective. Significant scatter in method performance was observed for the L1 benchmarks, reflecting, in part, a greater diversity of methods applied to the lowest-complexity benchmarks. For the higher complexity vehicle benchmarks, some stratification in general method performance was observed in the sea benchmarks with greater variation within assessment categories. All multi-fidelity methods performed poorly for the L3 air benchmark, potentially reflecting more challenging problem characteristics (e.g., physics-based constraints).
Date Issued
2025-03-18
Citation
Goal-Driven, Multi-Fidelity Approaches for Military Vehicle System-Level Design [STO-TR-AVT-331], 2025, pp.10-1-10-10
ISBN
978-92-837-2509-1
Publisher
NATO STO
Start Page
10-1
End Page
10-10
Journal / Book Title
Goal-Driven, Multi-Fidelity Approaches for Military Vehicle System-Level Design [STO-TR-AVT-331]
Copyright Statement
© 2025 NATO STO.
Identifier
https://publications.sto.nato.int/publications/STO%20Technical%20Reports/STO-TR-AVT-331/$$TR-AVT-331-ALL.pdf
Article Number
Chapter 10