A tool for the optimisation of vibro-acoustic systems using a parallel genetic algorithm and a distributed computing network

Abstract

An asynchronous parallel genetic algorithm (GA) was used in conjunction with a distributed computing network to optimise the locations and parameters for combined Helmholtz resonators and tuned mass dampers, called passive vibro-acoustic devices (PVADs), placed on the walls of a composite cylinder model of the payload bay in a rocket used to launch satellites. The vibro-acoustic response was calculated using modal coupling theory with the modal response calculated using commercial finite element software. Binary and integer representations of the chromosomes in the genetic algorithm were used in the optimisation, and it was found that the integer representation converged twice as fast as the binary representation. Optimisations were conducted with a varying number of PVADs. The use of the distributed computing environment reduced the time taken to conduct the optimisation to 3 days compared to 75 days on a single desktop computer.