In this paper a blind beamforming algorithm based on radial basis function neural network (RBFNN) is presented according to the array signal. The weight vector of beamforming is estimated by Doppler information of the signal first, then it is approximated by RBFNN to carry out the blind optimizing beamforming. This method not only can make the computation rapid but also is robust to system error because of the structural advantage that neural network can operate in parallel.The simulation experiment verifies the correctness of this algorithm.