Multi-objective Optimization of a Multi-chamber Perforated Muffler Using an Approximate Model and Genetic Algorithm

Perforated mufflers are widely used in automotive intake and exhaust systems and need to be properly designed. However, multi-objective optimization in practical perforated muffler designs usually involves finite element or boundary element models, which demand a higher computation time for evolutionary algorithms. In this paper, an approximate model for transmission loss (TL) predictions is established by correcting the thickness correction coefficient in the transfer matrix using the data calculated by the finite element model (FEM). The approximate model is computationally cheap and applicable for TL predictions above the plane wave cut-off frequency. A popular evolutionary algorithm, NSGA-, amalgamated with the approximate model, has been adopted to carry out the multi-objective optimization of a multi-chamber perforated muffler. The goals of optimization are to maximize TL at the target frequency range, as well as to minimize the valleys of TL and the size of the muffler. Both transmission loss and insertion loss of the optimized muffler are measured. Numerical and experimental results are in good agreement and show significant improvements of acoustic performance precisely at the target frequency range. Consequently, the combination of the approximate model and the NSGAalgorithm provides a fast, effective, and robust approach to co-axial perforated muffler optimization problems.