Node positioning is a basic but important research direction in wireless sensor networks. In practical applications, sensor nodes are mostly randomly deployed, and the distribution is often uneven. There is a low positioning accuracy of nodes in wireless sensor network iterative positioning algorithm. Aiming at the problem of traditional localization algorithm, a new localization algorithm is proposed based on genetic method in this paper. The algorithm uses the genetic model to obtain the node velocity and orientation information, then according to the node's historical trajectory of time series we can get accurate sampling region that can get the high quality sample points which is closer to the unknown node. Simulation results show that the performance of proposed algorithm is superior to the traditional algorithms.