Development of a multi-objective artificial tree (MOAT) algorithm and its application in acoustic metamaterials

Abstract

Although there are many algorithms that can solve the multi-objective optimization problems (MOPs) efficiently, each algorithm has its own disadvantages. The emergence of new algorithms is beneficial to make up the deficiencies of existing algorithms. Inspired by the organic matter transport process and the branch update theory of the banyan, this work proposed a new bio-inspired algorithm, named the multi-objective artificial tree (MOAT) algorithm to solve the MOPs. In MOAT, an improved crossover operator and an improved self-evolution operator are introduced to update solutions, a adaptive grid method is applied to manage the non-dominated solutions, and the strategy of variable number of branches in population is adopted to enhance the accuracy of this algorithm. Many typical test functions and seven well-known multi-objective algorithms, including MOEAD, NSGAII, MOPSO, GDE3, eMOEA, IBEA and MPSO/D, are applied to study the accuracy and efficiency of MOAT. Experimental tests show that the results of MOAT are better than those of the seven algorithms, and the performance of MOAT is demonstrated. In addition, this new algorithm is also applied to solve the MOPs of two-dimensional acoustic metamaterials (AMs). The key parameters of AMs are optimized by MOAT to mitigate impact load and reduce structural mass, and the performance of these AMs is significantly improved.