Abstract: A period-divided teaching-learning-based optimization is proposed for the bi-objective distributed heterogeneous hybrid flow shop scheduling problem with sequence-dependent setup times. Firstly, a three-direction initialization strategy based on the NEH heuristic is developed to enhance the quality of the initial population. Next, the optimization process is divided into two periods according to the number of iterations. The early period focuses on maintain population diversity, while the later period aims to accelerate convergence. In each period, different strategies are employed, including ruin-and-recreate, critical crossover search, differential mutation, neighborhood search, and others. Finally, an elimination strategy is implemented to ensure the overall convergence of the population. Comparative experiments on instances of various scales have demonstrated that the proposed algorithm achieves significant advantages in both convergence and diversity when compared with other algorithms.