TL;DR
This paper presents a real-time optimal guidance strategy for low-thrust spacecraft using Nonlinear Model Predictive Control (NMPC). By leveraging NMPC's ability to handle complex nonlinear dynamics and unmodeled disturbances, the authors demonstrate an efficient method for spacecraft orbit transfers, specifically from Low Earth Orbit (LEO) to a Molniya orbit, while minimizing fuel consumption.
What Problem Does This Paper Address?
Spacecraft guidance, especially for deep space missions, requires autonomous control due to communication delays. Low-thrust propulsion, while fuel-efficient, results in very long transfer times, making trajectory optimization challenging. Furthermore, spacecraft trajectories are subject to unmodeled disturbances (like gravitational effects, solar wind, and atmospheric drag) that can significantly alter the planned path. Traditional methods often rely on linearized models, which become inaccurate over extended horizons, or require complex, time-consuming computations. This paper addresses the need for a real-time, robust guidance system that can handle nonlinear dynamics and disturbances for low-thrust orbit transfers, optimizing fuel usage over long durations.
Methods
The core of the proposed solution is Nonlinear Model Predictive Control (NMPC). NMPC operates by repeatedly solving an optimal control problem (OCP) at each time step. A nonlinear model of the spacecraft's dynamics is used to predict future states, and an optimization algorithm computes the control actions needed to achieve a desired objective (e.g., reaching a target orbit) while satisfying constraints (e.g., fuel limits).
Key methodological aspects include:
- Direct Transcription: The continuous-time OCP is transformed into a finite-dimensional Nonlinear Programming (NLP) problem by discretizing the equations of motion and approximating state and control profiles using polynomials at collocation points. This approach avoids the difficulties of solving multi-point boundary value problems associated with indirect methods.
- Modified Equinoctial Elements: To represent the spacecraft's dynamics, the authors employ modified equinoctial elements. These elements are chosen over classical orbital elements or Cartesian coordinates because they avoid singularities for zero inclination and unitary eccentricity, making them more suitable for numerical simulations across a wide range of orbital conditions.
- IPOPT Solver: The resulting large-scale NLP is solved using IPOPT (Interior Point Optimizer), a primal-dual interior-point algorithm known for its efficiency in handling large nonlinear optimization problems.
Datasets
Not discussed in the paper. The study uses a simulated case study for evaluation.
Key Insights and Ideas
- NMPC for Long-Duration Transfers: The extended time scales of low-thrust orbit transfers are a significant advantage for NMPC. Unlike applications requiring near-instantaneous control, the long transfer times provide ample computational budget to solve the NMPC optimization problem at each step, making it a practical approach.
- Handling Unmodeled Disturbances: The NMPC framework inherently allows for the incorporation of unmodeled disturbances. By not including certain perturbing forces (like Earth oblateness) in the predictive model but accounting for them in the actual system implementation, the controller's ability to reject these disturbances can be effectively tested and demonstrated.
- Modified Equinoctial Elements for Robustness: The choice of modified equinoctial elements is crucial for avoiding numerical singularities that plague classical orbital elements, particularly for orbits with zero or near-zero inclination or eccentricity. This choice ensures a more robust and universally applicable dynamic model.
Key Results
The paper presents a case study of transferring a spacecraft from LEO to a Molniya orbit. The NMPC methodology successfully:
- Tracks the Optimal Trajectory: Figure 2 shows that the NMPC solution closely matches the trajectory computed by a full-time, precomputed optimal control solution (which assumes no disturbances).
- Minimizes Fuel Consumption: The methodology is designed to optimize thrust usage, leading to efficient fuel consumption for the transfer.
- Demonstrates Disturbance Rejection: By simulating the system with Earth oblateness effects while the NMPC controller's model omits them, the study shows the controller's ability to adapt and maintain the desired trajectory.
- Efficient Computation: Figure 3 indicates that the computational statistics (iterations, function evaluations) are very low compared to the overall transfer time, reinforcing the practicality of real-time implementation.
Limitations and Future Work
- Limitations:
- The paper acknowledges that while NMPC can handle unmodeled disturbances, a perfect model would still lead to a more optimal solution. The current study uses a simplified approach to demonstrate disturbance rejection.
- The robustness of the NMPC design under the presence of disturbances is an active research area, and while this paper provides a practical application, deeper theoretical analysis of robustness margins is suggested.
- Future Work:
- Further development of robust NMPC design to maintain stability under disturbances is encouraged.
- The paper suggests that the methodology presented is a promising strategy for real-time optimal guidance of low-thrust orbit transfers and encourages further development.
- The study highlights the potential for NMPC to handle complex nonlinear models and constraints, opening avenues for its application in other challenging space mission scenarios.
Summary updated