Optimizing Bio-Inspired Propulsion Systems using Genetic Algorithms

Abstract

Optimization is an important part of the development of an efficient propulsion system. The biomimetic marine propulsion system referred to as an oscillating foil has recently seen an upswing in interest. The optimization of movement patterns for these systems has yet to be researched in detail and has a potentially large payoff. Because of their biomimetic origins and problem type compatibility, using a genetic algorithm to optimize movement of oscillating foils is proposed. This thesis describes the construction of an experiment setup that implements a genetic algorithm to evolve a population using a physical environment to determine the fitness of its individuals. The experiment setup is a fully autonomous physical simulation of a real world application of a marine propulsion system, installed in the CWT-lab at NTNU. The usage of a physical environment is proven to be a quick and reliable way of calculating fitness for this application. Optimal movement patterns are found for most experiment cases. The results of the experiment show that optimizing oscillating foil movement patterns using genetic algorithms together with a physical test system is a viable option.