This paper investigates the aircraft maintenance task scheduling problem, integrating labor costs and uncertainty costs into a unified optimization framework. A multi-objective optimization method of maintenance task scheduling duration robustness based on NSGA-Ⅱ-Adaptive is proposed. The proposed method employs a novel approach combining critical path weights with buffer attenuation effects to define robustness values, and enhances the crossover and mutation processes of NSGA-Ⅱ through adaptive methods. The algorithm was tested on benchmark datasets of various scales and compared with NSGA-Ⅱs, NSGA-Ⅱ-LSA, and the standard NSGA-Ⅱ. Experimental results demonstrate that the proposed algorithm outperforms the counterparts in four performance metrics: convergence speed, number of non-dominated solutions, spacing metric, and coverage metric. Further validation through a real-world aircraft maintenance case study reveals that the algorithm reduces comprehensive deviations by 63%, 77.6%, and 77.9% compared to NSGA-Ⅱs, NSGA-Ⅱ-LSA, and NSGA-Ⅱ respectively. These findings confirm the proposed optimization method's practical applicability and effectiveness in addressing real-world maintenance scheduling challenges.