Supply requirement prediction during long duration space missions using Bayesian estimation

This paper examines the probabilistic relationship between resource consumption and crew workload in an analogue Mars Base scenario. We use data from the 2004 season of the Flashline Mars Arctic Research Station (FMARS) to define a probabilistic relationship between food consumption, planned workload, and actual work conducted by the crew. Bayesian estimation is then used as a mathematical method of learning this relationship. The learned model can be used as a basis for future logistics planning for a crew in a given environment—food supplies and work conducted would be tracked daily, allowing base mission operations to predict and adjust critical re-supply dates from learned data and a planned workload. We show results from field exercises, which demonstrate considerably greater prediction accuracy than current methods, and which are directly applicable to long-duration space missions, regardless of individual crew makeup and personal needs.