Study of Optimization Algorithms for Underwater Acoustic Applications

Abstract

This thesis introduces the algorithms harmony search (HS) and artificial bee colony (ABC) to the field of geoacoustic inversion. HS mimics jazz musicians search of an optimal harmony through improvisation, while ABC is inspired by bee?s search for food sources with a high amount of nectar. These global optimizers have been combined with downhill simplex (DS), a local optimizer, in order to create hybrid optimization algorithms. These hybrids have further been compared with a hybrid version of differential evolution (DE) by testing them on a problem of geoacoustic inversion. The goal was to see if these relatively recent algorithms could outperform the well known evolutionary algorithm. The hybrid based on HS is shown to have better performance than the DE-based hybrid, both in terms of computational cost and accuracy.