The position of UAV formation must be adjusted by positioning method to keep the formation during flight. Maintaining formation is conducive to completing various tasks of UAVs. In order to reduce external interference, electromagnetic signal emission should be reduced as much as possible. However, the existing passive location and active location methods can not effectively solve this problem. Active positioning will increase the risk of exposing the position of the UAV, and the UAV should not be too far away from the ground signal source. Existing passive localization methods such as building nonlinear motion model and volume Kalman wave algorithm cannot adjust the position of UAV quickly without reducing electromagnetic signal emission.To maintain formation shape, an optimization model based on azimuth-only passive positioning and greedy strategy is established. We studied a particular case of drone formation. Firstly, we give the location model by geometric method. Next, we build an optimization model based on greedy strategy. At last, We randomly selected a set of data for simulation, the UAV position can be adjusted only 6 times to achieve a better effect. It means fewer electromagnetic signals are emitted and the adjustment speed is faster.