Introduction
The aviation industry is undergoing a significant transition toward "more-electric" aircraft architectures. By replacing traditional hydraulic and pneumatic systems with electrical counterparts, manufacturers aim to reduce weight, improve fuel efficiency, and lower maintenance costs. However, this shift has led to an explosion in the complexity of Electric Power Systems (EPS). Designing these systems requires balancing stringent safety standards, extreme reliability, and optimal hardware utilization.
The Challenge of Modern EPS Design
Modern aircraft EPS must meet extraordinary reliability targets, often requiring a failure probability of less than $10^{-9}$ per flight hour. Furthermore, they must adhere to complex operational constraints, such as the "no-parallel" rule, which prevents two AC generators from being connected to the same bus simultaneously to avoid synchronization issues.
Traditionally, the industry has relied on heuristic, text-based requirements and incremental modifications to legacy designs. This manual approach is increasingly inadequate; it is prone to human error, often results in expensive late-stage redesigns, and makes it difficult to formally prove that a new, optimized configuration is safe.
Methodology: A Formal Design Flow
This paper proposes a rigorous, automated design methodology based on Platform-Based Design (PBD) and Assume-Guarantee Contracts. A contract formalizes the interface between system components: a component guarantees a specific behavior ($G$) provided its environment satisfies a set of assumptions ($A$). This abstraction allows for the decoupling of hardware topology from control software.
1. Topology Synthesis
The first stage focuses on the physical layout. The authors employ Mixed-Integer Linear Programming (MILP) to identify the lowest-cost interconnection of generators, buses, and switches. To handle the non-linear nature of reliability constraints, the framework uses an iterative loop. A candidate topology is generated via MILP, its reliability is assessed, and if it fails the $10^{-9}$ threshold, new linear constraints (cuts) are added back to the optimizer until a valid solution is found.
2. Control Synthesis
Once the topology is established, the system requirements are translated into Linear Temporal Logic (LTL). Using the TuLiP (Temporal Logic Planning) toolbox, the framework automatically synthesizes a state machine controller. Because this controller is derived directly from formal logic, it is "correct-by-construction," eliminating the need for extensive manual debugging of switching logic.
3. Real-Time Refinement
To bridge the gap between discrete logic and continuous-time physical performance (e.g., voltage transients during switching), LTL requirements are refined into Signal Temporal Logic (STL). The Breach toolbox is then used to monitor high-fidelity simulations, allowing designers to identify optimal timing parameters and ensure the system remains stable under dynamic conditions.
System Architecture and Evaluation
The methodology was evaluated using a representative aircraft EPS model adapted from a Honeywell patent. This model incorporates a variety of components, including generators, auxiliary power units (APUs), transformer rectifier units (TRUs), and batteries, organized into high-voltage and low-voltage subsystems.
Results and Discussion
The automated framework demonstrated significant improvements over manual design processes:
- Efficiency: Topology synthesis for a complex system was completed in approximately 20 seconds, while control synthesis took only 1.6 seconds. This represents a shift from weeks of manual effort to near-instantaneous generation.
- Safety and Reliability: The framework successfully synthesized multiple topologies that met the $10^{-9}$ reliability target.
- Performance Margins: Through STL-based simulation, the authors determined critical timing constraints. For instance, they identified that the controller must react within 6.5 ms during specific failure scenarios to maintain DC bus voltage within acceptable bounds.
Conclusion and Future Directions
By formalizing the design process through contracts and temporal logic, this methodology provides a path toward safer, more optimized aircraft power systems. While the current work focuses on centralized control, future research will explore scalability to distributed architectures and the use of richer formalisms to handle non-deterministic environment behaviors. Automating the search for worst-case failure scenarios in hybrid system models remains a key objective for enhancing the robustness of the design flow.
Summary updated