Abstract: For the NP-hard problem of construction task scheduling under complex logical constraints, a multi-objective mathematical model that included task coding constraints, material availability constraints, and process constraints was established. Under the premise of introducing external disturbances, the optimization objectives were to minimize the total project duration and maximize the coordination scheduling factor. To efficiently solve this robust multi-objective scheduling problem, a third-generation non-dominated sorting genetic algorithm (ECE-NSGA-III) incorporating an ecological community evolution mechanism was proposed. The algorithm adopted two-segment chromosome mapping encoding to describe task sequences and supply point selections, and employed nine evolutionary mechanisms to jointly enhance global and local search capabilities. Simulation experiments were carried out in non-traversable task scenarios. The results show that the proposed method outperforms traditional approaches in terms of solution quality and robustness, effectively improving construction task scheduling performance under disturbances.