As a crucial platform for integrating distributed renewable energy, the efficient and optimal scheduling of microgrid clusters is key to enhancing the economic efficiency and reliability of regional power supply. Addressing the high-dimensional, non-convex, and multi-constraint nature of the scheduling model, alongside the tendency of traditional optimization algorithms to trap in local optima, a collaborative optimization scheduling method for microgrid clusters based on an improved simulated annealing dragonfly algorithm (SADA) is proposed. First, a scheduling optimization model is established with the objective of minimizing the combined operational and environmental costs. Then, the logistic chaos mapping is employed to enhance initial population diversity, the nonlinear adaptive inertia weights are introduced to balance global and local search capabilities, and a simulated annealing mechanism is embedded to improve the ability to escape local optima by leveraging its probabilistic jump property. Finally, compared with other benchmark algorithms, the proposed SADA achieves reductions in total cost by 7.33%, 5.09%, and 4.24%, respectively, effectively validating its effectiveness and superiority in complex energy scheduling scenarios.