TL;DR
This paper introduces a novel approach to optimize aeroassisted vehicle trajectories by unifying multiple objectives and their priority constraints into a single-objective problem. This method leverages a Fuzzy Goal Programming (FGP) framework and a hybrid gradient-based evolutionary algorithm to efficiently generate feasible and high-quality flight paths, outperforming traditional multi-objective optimization techniques in handling prioritized requirements.
What Problem Does This Paper Address?
Designing optimal trajectories for aeroassisted vehicles is a complex challenge due to numerous uncertainties and constraints. Traditional approaches often focus on single objectives or struggle to effectively incorporate multiple, potentially conflicting, mission requirements with varying priorities. This paper addresses the need for a robust method that can simultaneously optimize several performance metrics (like minimizing time, aerodynamic heating, and maximizing fuel efficiency and final velocity) while strictly adhering to pre-defined priority levels for these objectives. Existing multi-objective evolutionary algorithms (MOEAs) can be computationally expensive due to sorting procedures and may not directly handle explicit priority constraints, leading to difficulties in achieving desired outcomes when certain objectives are more critical than others.
Methods
The core of this paper's contribution lies in its innovative optimization framework. It begins by transforming the multi-objective problem into a single-objective one using a Fuzzy Goal Programming (FGP) technique. This FGP approach fuzzifies the objective functions and incorporates priority constraints by defining "satisfactory degrees" and using fuzzy relations. This allows for the explicit representation of priority requirements, such as "objective A is more important than objective B."
The transformed single-objective problem is then solved using a custom-designed Gradient-based Hybrid Genetic Algorithm (GHGA). This algorithm enhances a standard Genetic Algorithm (GA) by incorporating a gradient-based local search operator. This hybrid approach aims to combine the global search capabilities of GAs with the faster convergence of gradient-based methods, leading to improved local search ability and overall efficiency. The constraint handling is managed through a "constraint violation degree" that prioritizes feasible solutions.
Datasets
Not discussed in the paper. The paper focuses on a simulated mission scenario for an aeroassisted vehicle, using defined vehicle dynamics, path constraints, and objective functions.
Key Insights and Ideas
- Unifying Priorities: The FGP framework is a powerful tool for explicitly embedding mission-critical priorities into the optimization problem, moving beyond simple Pareto optimality.
- Efficiency Boost: Transforming the multi-objective problem into a single-objective one eliminates the need for computationally intensive non-dominated sorting procedures common in traditional MOEAs, leading to significant speedups.
- Hybridization Power: Combining gradient-based local search with evolutionary algorithms offers a synergistic approach, leveraging the strengths of both to achieve better solutions more efficiently.
- Feasibility First: The constraint handling mechanism, which prioritizes individuals with lower violation degrees, ensures that the generated trajectories are not only optimal but also safe and physically achievable.
Key Results
The proposed FGP-GHGA algorithm demonstrated its effectiveness in generating feasible and high-quality aeroassisted vehicle trajectories across various priority scenarios. Simulations showed that the algorithm could successfully satisfy pre-assigned priority requirements, leading to distinct trajectory profiles depending on the defined priorities. For instance, prioritizing fuel efficiency (maximizing final mass) led to different control strategies compared to prioritizing speed or minimizing flight time. The paper also highlights that the GHGA algorithm achieved a good balance between global exploration and local exploitation, resulting in satisfactory performance across all tested cases. The average running time for the optimization was reported as approximately 6 hours and 19 minutes.
Limitations and Future Work
The paper acknowledges that the FGP-GHGA approach, while effective, relies on the definition of "satisfactory degrees" and "goal values" for objectives, which can be influenced by the designer's experience. The paper also notes that the augmented objective function (Jaug) might not be differentiable at all points, necessitating the use of smooth approximations for practical implementation.
Future work could involve exploring more advanced gradient estimation techniques or further refining the hybrid operator. Additionally, extending this framework to handle more complex mission scenarios with a larger number of objectives or more intricate priority structures could be a valuable direction. The authors also suggest that the FGP-GHGA solution can serve as a high-quality reference trajectory for online guidance systems.
Summary updated