Multi-variable optimisation of a vibro-acoustic system using a distributed computing network

Abstract

This paper describes a computational method involving the use of a genetic algorithm and a large number of networked desktop computers to optimise the locations and parameters of acoustic and vibration absorbers to reduce noise transmission into an enclosure. A Finite Element model of a vibro-acoustic system was created in ANSYS and the uncoupled structural and acoustic modal responses were calculated. The combined response of the acoustic and vibration absorbers, the acoustic cavity and structural modal responses were coupled using modal coupling theory in Matlab. Optimisations were conducted using a varying number of acoustic and vibration absorbers. The use of the networked desktop computers dramatically reduced the time taken to conduct the optimisation compared a single desktop computer. A factor of 25 improvement in speed was realised for the physical system examined.