Search Results

Advanced life support systems have many interacting processes and limited resources. Controlling and optimizing advanced life support systems presents unique challenges that are addressed in this paper. We have developed a controller using reinforcement learning [Barto&Sutton], which actively explores the space of possible control strategies, guided by rewards from a user specified long term objective function. We evaluated this controller using Biosim, our discrete event simulation of an advanced life support system. This simulation supports all life support modules including crew, food production, air revitalization, water recovery, solid waste incineration and power. Our algorithm for reinforcement learning discovered unobvious strategies for maximizing mission length. By exploiting nonlinearities in the simulation dynamics, the learned controller outperforms a controller designed by an expert.

This paper describes the role of transient simulations, heuristic techniques, and closed loop integrated control in designing and sizing habitat life support systems. The integration of these three elements allows for more accurate requirements to be derived in advance of hardware choices. As a test case, we used a typical lunar surface habitat. Large numbers of habitat configurations were rapidly tested and evaluated using automated decision support tools. Through this process, preliminary sizing for habitat life support systems were derived. Our preliminary results show that by using transient simulations and closed loop control, we substantially reduced the system mass required to meet mission goals. This has greater implications for general systems analyses and for life support systems.

BioSim is a simulation tool which captures many basic life support functions in an integrated simulation. Conventional analyses can not efficiently consider all possible life support system configurations. Heuristic approaches are a possible alternative. In an effort to demonstrate efficacy, a validating experiment was designed to compare the configurational optima discovered by heuristic approaches and an analytical approach. Thus far, it is clear that a genetic algorithm finds reasonable optima, although an improved fitness function is required. Further, despite a tight analytical fit to data, optimization produces disparate results which will require further validation.

This paper describes a simulation of an integrated advanced life support system. It contains models of the major components of an Advanced Life Support (ALS) system including crew, biomass, water recovery, air revitalization, food processing and power supply. The simulation also models malfunctions and stochastic processes. Sensors and actuators are modeled to allow controllers to interact with the simulation. The simulation is designed for testing and evaluation of life support control approaches. We use an example of a simple genetic algorithm to demonstrate how a control application might use the simulation. The simulation is implemented in Java to make it portable and easy to use.

The paper describes an approach to the integration of qualitative and quantitative modeling techniques for advanced life support (ALS) systems. Developing reliable control strategies that scale up to fully integrated life support systems requires augmenting quantitative models and control algorithms with the abstractions provided by qualitative, symbolic models and their associated high-level control strategies. This will allow for effective management of the combinatorics due to the integration of a large number of ALS subsystems. By focusing control actions at different levels of detail and reactivity we can use faster, simpler responses at the lowest level and predictive but complex responses at the higher levels of abstraction. In particular, methods from model-based planning and scheduling can provide effective resource management over long time periods.

Regenerative life support systems for long duration human space exploration missions present unique design challenges that are also reflected in constructing behavior models of these systems for analysis purposes. These systems have multiple modes of operation and complex non-linear dynamics that occur at multiple time scales. Coarse grained analysis of the complete system over long durations and fine grained temporal analysis of smaller system elements while avoiding computational intractability can be achieved by using multiple modeling and simulation paradigms. We present a multi-level simulation model of an advanced life support system. The simulation model couples a discrete-event approach at the system level, with more detailed hybrid (continuous/discrete) physical system modeling at the sub-system level.

High costs and extreme risks prevent the life testing of NASA hardware. These unavoidable limitations prevent the determination of sound reliability bounds for NASA hardware; thus the true risk assumed in future missions is unclear. A simulation infrastructure for determining these risks is developed in a configurable format here. Positive preliminary results in preparation for validation testing are reported. A stochastic filter simulates non-deterministic output from the various unit processes. A maintenance and repair module has been implemented with several levels of complexity. Two life testing approaches have been proposed for use in future model validation.

A Lunar habitat will be highly sensored and generate large amounts of data or telemetry. For this data to be useful to humans monitoring these systems and to automated algorithms controlling these systems it will need to be converted into more abstract data. This abstracted data will reflect the trends, states and characteristics of the systems and their environments. Currently this data abstraction process is manual and ad hoc. We are developing a Data Abstraction Architecture (DAA) that allows engineers to design software processes that iteratively convert habitat data into higher and higher levels of abstraction. The DAA is a series of mathematical or logical transformations of telemetry data to provide appropriate inputs from a hardware system to a hardware system controller, system engineer, or crew. The DAA also formalizes the relationships between data and control and the relationships between the data themselves.

A Lunar habitat will require a level of automation that is much greater than previous human space missions. The complexity of the habitat, the distance (and time delay) between the habitat and ground controllers and the fact that the habitat may be uncrewed for periods of time all point towards increased automation of the habitat. NASA JSC is developing an integrated testbed for exploring operational concepts for a Lunar habitat that includes significant automation. The testbed allows for early investigation of the hardware and software design decisions and their impacts on operating a Lunar habitat. The testbed also allows for investigation into the robustness of different automation concepts with respect to failures and perturbations of the system. The testbed consists of both dynamic simulations of habitat systems and some physical hardware-in-the-loop.