An enhanced shuffled Shepherd Optimization Algorithm for optimal design of large-scale space structures

Abstract

Shuffled Shepherd Optimization Algorithm (SSAO) is a swarm intelligence-based optimizer inspired by the herding behavior of shepherds in nature. SSOA may suffer from some shortcomings, including being trapped in a local optimum and starting from a random population without prior knowledge. This study aims to enhance the performance of the SSOA by incorporating two efficient devices. The first device is utilized from the Opposition-Based Learning (OBL) approach to improve the initialization phase of the algorithm. The second device is incorporated a solution generator in the cyclic body of the SSOA based on the statistical results of the solutions. This feature is the so-called statistically regenerated stepsize. The proposed devices provide a good balance between exploration and exploitation capability of the algorithm and reduce the probability of getting tapped in a local optimum. The viability of the proposed Enhanced Shuffled Shepherd Optimization Algorithm (ESSOA) is demonstrated through three large-scale design examples. ESSOA is compared to the standard SSOA and some other existing metaheuristic algorithms. The optimization results reveal the competence and robustness of the ESSOA for optimal design of the large-scale space structures.