Memory-assisted adaptive multi-verse optimizer and its application in structural shape and size optimization

In this paper, an enhanced, memory-assisted version of the multi-verse optimizer (MVO) is proposed for the shape and size optimization of various types of truss structures. The cosmology-based MVO processes the interaction of the multiple universes through white, black, and wormholes to solve the target optimization problem. However, the MVO has drawbacks such as slow convergence speed, easy to fall into local optimum, and low efficiency in the case of continuous and discrete variables existing in structural optimization. This paper aims to modify the Multi-verse Optimizer and propose an enhanced version, named as memory-assisted adaptive multi-verse optimizer (MAMVO). The enhanced algorithm seeks to reinforce the performance of the standard algorithm using two mechanisms: (I) Using an adaptive pseudorandom transfer strategy in cases with both continuous and discrete variables to strike a better balance between the diversification and the intensification tasks (II) using a multi-elite memory, by which several best so far solutions are saved and exchanged with several worst ones for better convergence. The performance of the proposed MAMVO is compared with those of the standard MVO and some other optimization methods through various types of truss benchmarks (beams, towers, and bridges) under different constraints such as displacement, stress, buckling, or frequency. Results confirm that MAMVO significantly outperforms the standard MVO and has comparable or superior performance to the other methods.