Improved team learning-based grey wolf optimizer for optimization tasks and engineering problems

Optimization refers to finding the optimal solution to minimize or maximize the objective function. In the field of engineering, this plays an important role in designing parameters and reducing manufacturing costs. Meta-heuristics such as the grey wolf optimizer (GWO) are efficient ways to solve optimization problems. However, the GWO suffers from premature convergence or low accuracy. In this study, a team learning-based grey wolf optimizer (TLGWO), which consists of two strategies, is proposed to overcome these shortcomings. The neighbor learning strategy introduces the influence of neighbors to improve the local search ability, whereas the random learning strategy provides new search directions to enhance global exploration. Four engineering problems with constraints and 21 benchmark functions were employed to verify the competitiveness of the TLGWO. The test results were compared with three derivatives of the GWO and nine other state-of-the-art algorithms. Furthermore, the experimental results were analyzed using the Friedman and mean absolute error statistical tests. The results show that the proposed TLGWO can provide superior solutions to the compared algorithms on most optimization tasks and solve engineering problems with constraints.