关键词: 高维复杂函数优化, 灰狼优化, 反向学习, 醉汉漫步, CEC2013, 运算放大器

Abstract: The grey wolf optimization algorithm is easy to fall into the local optimum in the later stage of optimization. Because of its higher complexity when solving high-dimensional functions, the probability of falling into the local optimum is greater. To address this problem, this paper proposes a mixed grey wolf algorithm on the basis of both drunkard strolling and reverse learning, termed as DGWO. In the process of iteration, the dominant wolves are partially retained from the comparison between the dominant and the worst wolves via backward learning. Meanwhile, drunkard strolling is performed on the leader wolf, where the coefficient scalars are utilized in A and C instead of the coefficient vectors in the original algorithm. The effectiveness of the proposed method is investigated by 10 standard test functions (100D, 500D and 1 000D) as well as 10D CEC2013 test function, and compared with PSO, GWO-CS, and GWO algorithms. The simulation results demonstrate that the proposed DGWO algorithm performs better in terms of accuracy and convergence rate. In addition, the improved grey wolf algorithm is applied to the parameter design of the two-stage operational amplifier with the goal of maximizing the open-loop low-frequency gain to verify the practicability of our scheme.

Key words: high dimensional complex function optimization, grey wolf optimizer, reverse learning, drunkard strolling, CEC2013, operational amplifier