A genetic algorithm approach to optimal sensor placement in liquid hydrogen fuel tanks

Abstract

Genetic Algorithms (GAs) offer a robust framework for optimization when vast configuration spaces render analytical methods computationally infeasible. This research employs GAs to determine optimal fill-level sensor configurations for liquid hydrogen (LH2) aircraft tanks, addressing a gap in current aerospace instrumentation design. A pivotal element of this approach is the design of the fitness function, which quantifies system mass and sensor sensitivity across the full range of fill levels and operational flight envelopes (pitch and roll). To evaluate candidate configurations, the fluid volume content of the tilted tank and the intersection of the sensors with the fuel surface are computed. The results demonstrate that GAs can evolve novel, high-performance, and non-intuitive sensor layouts that fulfill operational requirements. However, the findings also underscore that integrating domain-specific constraints into the fitness operators is essential to ensure computationally efficient convergence toward physically viable solutions.