Unmanned aerial vehicles with the characteristics of zero casualties, low cost and good mobility, have broad application prospects in modern warfare and civilian areas. Conflict detection technology of UAV in an uncertain environment is prerequisite that multi-UAVs finish a task in the same airspace. In this paper, with unknown wind condition, the joint estimation of both the path and unknown parameter is presented in the risk of conflict detection algorithm. First, in the two-dimensional Cartesian coordinate system, the aircraft kinematic model under the influence of unknown wind is established, a particle learning filter method to estimated the flight track and the wind speed vector. Secondly, with the minimum horizontal and vertical safe distance, the collision risk probability of two planes in three-dimensional space is established and calculated. Finally, with a numerical example based on Matlab, the probability of collision risk is calculated, which verified the model is reasonable.