The invention discloses a K mean value clustering iteration based UAV sound source positioning method. According to the invention, on a basis of calculating time delay information of multiple groups of sound sensors by utilizing a generalized cross-correlation function, a time delay relation is utilized for writing various different linear system of equations and a plurality of UAV sound source three-dimensional coordinates are calculated out; then a K mean value clustering algorithm is utilized for estimating the optimal position coordinates of the UAV; and finally an iteration algorithm is adopted for calculating the final position coordinates of the UAV by taking the position coordinates obtained by using the clustering algorithm. According to the invention, time delay can be utilized fully for estimating the redundant information; through combination with advantages of the multiple sound sensors, the position of the UAV sound source object can be positioned more accurately and reliably; a shortcoming of sensitivity to initial values in the iteration algorithm is made up and a result solved through the linear system of equations is utilized as the initial value of iteration, sothat convergence rate is improved and the accuracy of a convergence result is ensured at the same time. The method provided by the invention has great application value.