This thesis report describes an optimisation procedure, applied to gearboxes, based on a multi-objective evolutionary algorithm (NSGA II®). The weight of a two stage gearbox and a second objective function (Noise Power Index), used to estimate the noise emitted, were minimized. A developed software tool (NaWOST) was used to perform this optimisation. A FEM analysis of the gearbox housing was automated by API called from Matlab®. The dynamic behaviour of the gears was simulated using a simplified lumped parameter model in a Simulink® environment. Then, the main harmonic components of gears meshing forces were compared with the natural frequencies, given by the FEM model, in order to estimate the resonance risk and subsequently to evaluate the noise objective function. Sound A-ponderation curve and normal mode shapes of the housing were also used to evaluate the emitted noise.
This thesis work was developed at the Manufacturing Department of the Cranfield University (UK).