This thesis deals with the application of reinforcement learning techniques to the control of a closed life support system simulator, such as could be used on a long duration space mission. We apply reinforcement learning to two different aspects of the simulator, control of recycling subsystems, and control of crop planting schedules. Comparisons ...