Design Methodology for Optimal Power Generation in Solar Vehicles Using Genetic Algorithms
This JHU/APL invention applies genetic algorithms to the problem of optimally determining solar panel location and size on an airship (or other solar paneled vehicle) to minimize power, weight, and volume for significant performance advantages over more “intuitive” design methods typically employed.
Genetic algorithms are search algorithms based on the mechanics of natural selection and propagation of superior genes. In a general sense, these algorithms encode potential solutions to a specific problem into the simple chromosome-like data structures of individuals of a population. The population is then subjected to a probabilistic “survival of the fittest” selection process to determine breeding pairs, wherein “fitness” is determined by the fitness of the encoded solutions to the subject problem. Mating is modeled as a genetic splice of parent individuals to form offspring to populate a new generation. Modeling of this genetic crossover includes a randomization of splice point(s) including a probability of clones and genetic mutations. The fitness of the members of the new population is then evaluated. If the genetically encoded solution of the fittest individual is inadequate, the selection and breeding processes are repeated to generate subsequent generations until a satisfactory solution is obtained. To improve the efficacy of optimization, “elitism” is often used to guarantee that the fittest individual is cloned into the next generation so the best solution at the time is not lost. The primary design challenge of reducing weight and volume of the vehicle is selecting the optimum solar panel locations (and hence minimum solar cell subsystem weight) which drives power, mass and volume. This challenge is present in virtually all solar-powered lighter-than-air vehicles and spacecraft employing body mounted panels, and it is traditionally satisfied in the following way: 1. Fully populate the surface to guarantee greatest possible power production in any attitude (at the expense of mass, thermal complications, integration and testing complications and other undesirable interactions with other subsystems. 2. Simplistically populate “obvious” surfaces (resulting in potentially significant operational constraints and/or additional attitude control requirements, but easier performance analyses). This JHU/APL prototype uses genetic algorithms to optimally determine solar panel location resulting in mass and volume reduction while allowing aggressive performance requirements. It results in demonstrably superior designs and can be readily adapted to optimizing spacecraft using body-mounted panels or fixed deployed panels. The invention offers even more significant advantage to emergent nanosat and picosat spacecraft whose mass, volume, and power requirements are exceedingly constrained and whose attitude control capabilities are generally limited.
*Seeking exclusive licensee.CONTACT: