An improved optimization algorithm to search for an optimal solution of a muffler design problem

Abstract

In this work, an auxiliary algorithm is proposed to assist a gradient-based optimizer in searching for an optimal solution of a muffler design problem formulated by topology optimization. The proposed algorithm escapes an objective function from one of local minima and pushes it toward a global optimum. Its key idea is adjusting a sign of sensitivity, the first derivative of an objective function with respect to a design variable, by multiplying it with 1 or -1. In an acoustical topology optimization problem formulated for muffler design, the absolute value of an acoustic pressure at the outlet is selected as an objective function, and the number of allowed finite elements for partitions inside the muffler is constrained. The formulated muffler design problem is solved for several design conditions, and the numerical results support the validity of the proposed auxiliary algorithm.