An Efficient Growth Optimizer with Adaptive Parameters and Targeted Stochastic Mutation Strategies for Global Optimization

Growth optimizer (GO) is a metaheuristic algorithm that simulates the learning and internal self-reflection mechanisms during human growth. The algorithm exhibits stable and efficient convergence capabilities on mathematical bench-mark functions of various types and complexities. However, it still suffers from trapping into local optimal solution. Thus, in this paper, we propose an improved growth optimizer algorithm named ASGO. The features of ASGO are summarized as follows: first, it adopts a dynamic adjustment mechanism combining individual learning and reflection speeds, and achieves adaptive adjustment during iteration through decay factors. Second, in the reflection phase, the targeted random mutation strategies is introduced to increase population diversity. The results showed that ASGO can reduce the premature convergence and enhance the algorithm’s global search efficiency and adaptability. To verify the effectiveness of the proposed method, the comprehensive experiments were con-ducted on 30 benchmark functions of CEC2017 and compared against nearly 50 popular efficient metaheuristic algorithms, and the statistical results confirm the high performance of the ASGO algorithm.