Uncertainty and Error in Combat Modeling, Simulation, and Analysis

Abstract

Due to the infrequent and competitive nature of combat, several challenges present themselves when developing a predictive simulation. First, there is limited data with which to validate such analysis tools. Secondly, there are many aspects of combat modeling that are highly uncertain and not knowable. This research develops a comprehensive set of techniques for the treatment of uncertainty and error in combat modeling and simulation analysis. First, Evidence Theory is demonstrated as a framework for representing epistemic uncertainty in combat modeling output. Next, a novel method for sensitivity analysis of uncertainty in Evidence Theory is developed. This sensitivity analysis method generates marginal cumulative plausibility functions (CPFs) and cumulative belief functions (CBFs) and prioritizes the contribution of each factor by the Wasserstein distance (also known as the Kantorovich or Earth Movers distance) between the CBF and CPF. Using this method, a rank ordering of the simulation input factors can be produced with respect to uncertainty. Lastly, a procedure for prioritizing the impact of modeling choices on simulation output uncertainty in settings where multiple models are employed is developed. This analysis provides insight into the overall sensitivities of the system with respect to multiple modeling choices.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 19, 2019
Accession Number
AD1089578

Entities

People

  • Jason A. Blake

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Advanced Electronics
  • Energy and Power Technologies
  • Human Systems

DTIC Thesaurus Topics

  • Air Force
  • Artificial Neural Networks
  • Bayesian Networks
  • Climate Change
  • Combat Simulations
  • Computational Science
  • Computer Simulations
  • Data Science
  • Engineering
  • Experimental Design
  • Information Science
  • Materials
  • Materials Engineering
  • Materials Science
  • Mathematical Models
  • Network Science
  • Operations Research
  • Predictive Modeling
  • Probability
  • Reliability
  • Simulations
  • Standards
  • Surveys
  • Test And Evaluation
  • United States
  • Virtual Reality

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Computational Modeling and Simulation
  • Systems Analysis and Design