Optimization of multi-structural parameters in metamaterials based on the DGN co-simulation method

The convergence of algorithms is an unavoidable problem when using global optimization algorithms to optimize acoustic properties of metamaterials. The quality of optimization of local optimization algorithms is often limited by the initial data. Moreover, the influence of structural parameters on the performance is difficult to be reflected in the optimization process of traditional algorithms. Thus, a combination algorithm optimization strategy for metamaterials in terms of multiple structural parameters is proposed in this paper based on a co-simulation approach. This strategy combines the design of experiments (DOE), genetic algorithm (GA), and NLPQL algorithm, which is referred to as the DGN method. For the optimization problem of complex structures, firstly, the relationship between the structural parameters on acoustic performance can be obtained by fitting the relationship between design factors and the response function through DOE. Then the global algorithm is combined with the local algorithm to solve the problem of poor convergence of the global optimization algorithm while ensuring the optimization quality of the local optimization algorithm. Compared with the original structure, the optimized metamaterial structure has an optimization effect of 44.8% for the peak frequency position of sound insulation as well as an optimization effect of nearly 116.7% for the bandwidth of sound insulation. Compared with the optimization effect of single algorithm (NSGA-II), this method improves the optimization effect of acoustic isolation bandwidth by 36.8%. The optimized structure reflects better low-frequency sound insulation performance. Therefore, this optimization method provides a new idea for the design and performance regulation of metamaterials.