To improve the accuracy of passive positioning technology and its environmental adaptability, a new nonlinear filter method Extend Square-root Cubature Kalman Filter(SCKF) is offered for multi-station passive location with three moving angle-measured sensors’ measurements. Firstly, Empirical Mode Decomposition (EMD) algorithm is used to estimate measurement noise covariance. And then the covariance of the procession noise and measurement noise is brought into the circle procession; At the same time, Cubature Kalman filter (CKF) is improved by the way of square root to keep the stability and positivity, and the results of tracking by Extend SCKF are compared with the results by Unscented Kalman Filter (UKF). By the tracking results to the velocity of the target, Extend SCKF algorithm can not only track the target with unknown measurement noise, but also improve the passive position precision remarkably with the same complexity to UKF.