Reinforcement Learned Adversarial Agent (ReLAA) for Active Fault Detection and Prediction in Space Habitats

Abstract

With growing interest for human space travel in the 21st century, much attention has been directed to the robust engineering of life sustainment systems in deep space habitats. The stable, reliable operation of such a habitat is partly achieved with an ability to recognize and predict faults. For these two purposes, a reinforcement learning adversarial agent (ReLAA) that is trained with experimental data to actively recognize and predict faults is introduced in this work. ReLAA proposes activating known faults in a system then simulates a digital twin of the mock space habitat to discover vulnerabilities which would lead to faulted operation. The methods developed in this work will allow for the discovery of damaging latent behavior, and the reduction of false positive and negative fault identification.