The invention discloses a reservoir multi-target water supply capacity optimization method based on a particle swarm optimization algorithm. The reservoir multi-target water supply capacity optimization method comprises the following steps: S1, randomly generating a primary agricultural scheduling line; S2, obtaining the water supply amount and the guarantee rate of the reservoir for each water user; S3, preliminarily selecting a primary agricultural scheduling line; S4, acquiring a water supply capacity range of the selected primary agricultural dispatching line; S5, obtaining the maximum water supply capacity of the selected primary agricultural dispatching line; S6, obtaining an individual optimal value and a global optimal value of the agricultural scheduling line; S7, optimizing the primary agricultural scheduling line by updating the individual optimal value and the global optimal value; and S8, taking the agricultural scheduling line which has the same water supply capacity as the agricultural scheduling line optimized last time and continuously reaches a preset algebra as an output. According to the invention, the maximum water supply capacity meeting the requirements of double assurance rates of certain agricultural irrigation water and urban water can be obtained, and the problem that multi-target multi-assurance-rate reservoir water supply cannot be processed in theprior art is solved.