Abstract: Abstract：Objectives The notable feature of the island power supply system is that it contains a wide variety of distributed power sources with complex topology, which can generate power independently from the main network, avoid the traditional direct connection to the main network, and effectively reduce network loss, voltage fluctuation, laying difficulty and maintenance cost. Because of its strong pheromone dynamic response ability, the adaptive ant colony algorithm has the ability to search the global optimal solution accurately and quickly, and will not fall into the local optimal. Methods Based on this, the capacity of the reef hybrid power generation system is optimized based on adaptive ant colony algorithm to maximize the use of renewable energy and adapt to the "two-carbon" policy. Results Taking Wailingding Island as the target island, a hybrid power generation system model of "Fengfengchai Storage" was built, and its capacity was optimized by adaptive ant colony algorithm. Conclusions The simulation results of this algorithm show that compared with the improved grey Wolf algorithm and artificial bee colony algorithm, the adaptive ant colony algorithm can effectively reduce the operating cost and environmental pollution of the hybrid power generation system of microgrid, and ensure the stability of power supply.