Finite-Horizon Robustness: Moving Beyond Traditional Stability Analysis

Abstract

This proposal aims to develop theoretical and computational tools to assess the robustness of systems over finite time horizons. A motivation for this work is the reliable assessment of multivehicle interceptor system performance, which will also enable robust design of such systems. Typical performance metrics are infinite-horizon in nature, centered on stability, and rely on frequency-domain concepts, such as gain/phase margins. Such metrics can be inadequate forsystems operating on finite-time horizons, as in many launch scenarios. Instead, this proposal focuses on time-domain metrics, e.g. bounds on the state of the system (tracking error, etc.) at the final and intermediate times of the horizon while considering the impact of disturbances, model uncertainty/variability, and initial conditions. The proposed work is organized around three complementary thrusts: 1) Finite-time robustness analysis for uncertain linear time-varying models, 2) Finite-time reachability analysis for uncertain nonlinear models, and 3) Software and naval applications. The goal in Thrust 1 is to compute robust bounds on the system state at intermediate and final times with the help of time domain dissipation inequalities that enable efficient solutions with convex optimization tools.The goal in Thrust 2 is to develop a scalable reachability method that takes advantage of numerical simulations and provide bounds on nearby trajectories that are not specifically simulated to over-approximate the reachable set. The goal in Thrust 3 is to develop software to perform the two proposed types of finite-horizon analyses and to apply the results to large-scale 3-d instances of multiple-shot interceptor problems.

Document Details

Document Type
DoD Grant Award
Publication Date
Jul 10, 2018
Source ID
N000141812209

Entities

People

  • Andrew Packard

Organizations

  • Office of Naval Research
  • United States Navy
  • University of California Regents

Tags

Fields of Study

  • Computer science

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Distributed Systems and Data Platform Development