Design optimization of an accumulator for reducing rotary compressor noise

This article deals with the optimization of an accumulator to reduce rotary compressor noise by maximizing the total transmission loss of an accumulator subjected to constraints on lower bounds of local transmission loss values. Transmission loss of an accumulator is described using the Helmholtz resonator theory to obtain the compressor noise characteristics. An experimental analysis of an accumulator is performed to identify the noise characteristics, and numerical data for use in design optimization are then obtained based on the Helmholtz resonator theory. Dominant design variables are selected via analyses of means, and are verified by back-propagation neural network based causality analysis. The back-propagation neural network is used to establish global approximate meta-models to accommodate the inherent non-linearity and multi-modality in the accumulator model. The genetic algorithm is employed as a global optimizer to find optimized accumulator designs in the context of approximation optimization.