<p>The quantum approximate optimization algorithm (QAOA) plays a crucial role in enhancing the performance of near-term noisy intermediate-scale quantum (NISQ) devices. This algorithm optimizes parameterized quantum circuits using classical optimizers to tackle complex combinatorial optimization problems.Addressing issues such as low quantum resource utilization, sensitivity to sampling noise, and complexity in obtaining optimal parameters for QAOA circuits using current optimizers, we propose an improved adaptive quantum genetic algorithm (IAQGA) as a gradient-free classical optimizer for QAOA. This optimizer employs a novel adaptive angle adjustment strategy that dynamically adjusts rotation angles based on the iteration count and cosine similarity, enhancing global search capabilities and solution quality. Performance evaluation against five state-of-the-art optimizers on the MaxCut problem demonstrates that our proposed optimizer significantly enhances the QAOA success probability and approximation ratios, showing statistically significant improvements. This validates the feasibility of our proposed optimizer.</p>