Abstract: A kind of underground positioning algorithm based on particle swarm optimization-generalized regression neural network(PSO-GRNN) is proposed.The proposed PSO-GRNN algorithm builds underground positioning model by the fast learning speed and strong approximation ability of generalized regression neural network(GRNN) and adjusted GRNN's smoothing parameters by using particle swarm optimization algorithm(PSO) to reduce the impact of human factors on selecting GRNN smoothing parameters to minimum.Finally, the coordinates of unknown nodes of underground can be directly obtained from the output of PSO-GRNN model.Simulating results show that the positioning accuracy of the PSO-GRNN model is better than that of the GRNN model and the BP model, and the algorithm complexity is lower and the efficiency is higher than that of the BP model, and meets the requirement of adaptive underground positioning.