Nested Multistate Design for Maximizing Probabilistic Performance in Persistent Observation Campaigns

Abstract

The paper presents a nested multistate methodology for the design of mechanical systems (e.g., a fleet of vehicles) involved in extended campaigns of persistent surveillance. It uses multidisciplinary systems analysis and behavioral-Markov modeling to account for stochastic metrics such as reliability and availability across multiple levels of system performance. The effects of probabilistic failure states at the vehicle level are propagated to mission operations at the campaign level by nesting various layers of Markov and estimated-Markov models. A key attribute is that the designer can then quantify the impact of physical changes in the vehicle, even those physical changes not related to component failure rates, on the predicted chance of maintaining campaign operations above a particular success threshold. The methodology is demonstrated on the design of an unmanned aircraft for an ice surveillance mission requiring omnipresence over Antarctica. Probabilistic results are verified with Monte Carlo analysis and show that even aircraft design parameters not directly related to component failure rates have a significant impact on the number of aircraft lost and missions aborted over the course of the campaign.