Solar Sails

Solar sails have been proposed as a cost effective source of space propulsion for a variety of future space exploration missions. Solar sails gain momentum from incident and reflected photons, and the continuous sunlight pressure provides sufficient propulsive energy for space missions that, otherwise, is only possible with a significant amount of propellant for conventional rocket systems. Currently, solar sail technology is being developed by the In-Space Propulsion Technology Program, managed by [NASA's Science Mission Directorate][43] and implemented by the In-Space Propulsion Technology Project at [Glenn Research Center][44]. The program's objective is to develop in-space propulsion technologies that enable NASA space science missions by significantly reducing cost, mass and travel times.

Research Objective

In the development of an appropriate control method, major challenges are associated with the uncertainties inherent in flexible solar sails because a comprehensive test for structural analysis is not possible in ground tests due to gravity on Earth. Even when the vacuum and thermal conditions of the space environment are well simulated, solar sail tests must employ awkward gravity offload systems to mitigate the effects of gravity. Further uncertainties in the material properties, test conditions, and modelling errors make it extremely difficult to obtain accurate flexibility characteristics of a flexible solar sail. Therefore, it is highly desirable for a control system to be able to adapt and compensate for system uncertainties. In our effort, we conceptually address the problem of flexibility from a perspective that is different from those in the literature. Prior to the commencement of a solar sail mission, the packaged sail must be deployed into its operational configuration. Maintaining stable attitude dynamics of the sailcraft/bus system will be a challenge due to the enormous increase in sailcraft inertia as the support structure and membranes deploy. Therefore, our tasks include conceptual modeling for solar sail deployment, control algorithm during and after deployment. Since, solar sail membranes are extremely difficult to characterize, in this study we focus on the behavior of booms that support membranes in solar sails.

Tasks

Simulations with NN-based control algorithm during deployment

In order to avoid excessive complexities related to the modelling of a deploying flexible solar sail, we consider a growing single boom that supports the solar sail membrane and further simplify it as a double flexible pendulum. Two masses evolve into its final configuration, mimicking the growth of the supporting boom of the structure, and we address how NN-based control system adapts to those adversary uncertainties during and after deployment. For further simplicity, the central hub from which the sail boom emerges is assumed fixed. Control system is also assumed as evolving, in which a neural network is added to augment a lead controller in each pendulum.

Experiments with the 30m SAFE boom

This structure was previously used in a Space Shuttle mission, and it has now been set up for control-structure interaction studies at the NASA Marshall Space Flight Center. In this study, three pairs of collocated accelerometers and air-jet thrusters mounted at the tip of the boom are utilized to suppress unwanted vibrations. The experiment is carried out by considering three-dimensional motions of the boom. As a result of long mission times, the boom may suffer gradual damage that results in inelastic deformation. For example, an inelastic bow in the boom will couple the bending and torsion at the tip where accelerometers and thrusters are mounted. In this research, instead of going through this process, we illustrate a NN-based adaptive control design in which a previous linear control design approach, which performed poorly on the current structure, is augmented to account for modeling uncertainty. A linear controller is designed assuming that bending in the X-Y directions are decoupled from each other as well as from torsion. Two proportional-integral (PI) controllers are designed identically assuming identical modal properties in the X-Y directions. This greatly simplifies the design procedure compared to designing a single controller for the coupled system of dynamics that is not available in our study. Separate but identical NNs are added to compensate for structural uncertainties.
Since the PI controllers are intended to control only bending motion in a single direction and the NNs are implemented independently, the overall design follows a decentralized architecture.

Stabilization after initial disturbances

Continuous external disturbances

Collaborators

The research was supported by NASA Marshall Space Flight Center, and Jerry Oakley and Mike Law at NASA Marshall set up the experiment which include integrating a Simulink-based controller with National Instruments Labview module. The tests shown in the above movies are carried out by Jerry Oakley.

Related links:

Presentations