Masters Category - Flight Test Point Optimization Program For a Self-Protection Application

Start Date

5-4-2021 10:00 AM

End Date

5-4-2021 10:30 AM

Document Type

Presentation

Abstract

Planning test points for highly constrained flight tests is a lengthy and iterative process which requires a structured, methodical approach using scientific test and analysis techniques (STAT) including: design theory, multi-objective optimization, and uncertainty analysis. Flight test engineers can spend anywhere from a week to a month determining ideal test points, only to have unforeseen problems arise during the week of testing that can invalidate these points. Genetic algorithms can play a key role in point selection for two common types of tests: model verification and validation (V&V) testing and operational test (OT) design. This paper outlines the methodology behind building a program to quickly identify a set of optimal test points for the trade space. The tool will allow test planners to have confidence in their test point design prior to the test as well as to make on-the-fly adjustments to testing locations during the event based on actual performance. There are a wide variety of parameters captured in the overall evaluation criteria (OEC) that give the planner great flexibility in tailoring the genetic algorithm outcomes for their purpose. The paper will begin by going through the steps behind planning a test and in defining the trade space and underlying uncertainty. Then, it will cover the parameters of the genetic algorithm and future work and recommendations.

Share

COinS
 
Apr 5th, 10:00 AM Apr 5th, 10:30 AM

Masters Category - Flight Test Point Optimization Program For a Self-Protection Application

Planning test points for highly constrained flight tests is a lengthy and iterative process which requires a structured, methodical approach using scientific test and analysis techniques (STAT) including: design theory, multi-objective optimization, and uncertainty analysis. Flight test engineers can spend anywhere from a week to a month determining ideal test points, only to have unforeseen problems arise during the week of testing that can invalidate these points. Genetic algorithms can play a key role in point selection for two common types of tests: model verification and validation (V&V) testing and operational test (OT) design. This paper outlines the methodology behind building a program to quickly identify a set of optimal test points for the trade space. The tool will allow test planners to have confidence in their test point design prior to the test as well as to make on-the-fly adjustments to testing locations during the event based on actual performance. There are a wide variety of parameters captured in the overall evaluation criteria (OEC) that give the planner great flexibility in tailoring the genetic algorithm outcomes for their purpose. The paper will begin by going through the steps behind planning a test and in defining the trade space and underlying uncertainty. Then, it will cover the parameters of the genetic algorithm and future work and recommendations.