To solve the problems of estimation accuracy restricted by large initial estimation error and unknown noise statistics in bearings-only tracking

a divided differential filter with intelligent statistical noise estimator was proposed. An S-H intelligent noise statistical estimator was proposed according to statistical linear regression theories

and was used for optimizing measurement update step of traditional divided differential filter

thus unknown state and noise measurement were intelligently and statistically calculated. The ability of the filter to adapt to complex nonlinear functions was further improved by iteration updating. Results showed that for typical passive tracking problem in linear state function and nonlinear measurement function with relative large initial estimation errors

the proposed filter provided better performance of nonlinear estimation task compared to several mainstream adaptive filters when the statistical characteristics of the system noise and measurement noise were unknown

and it effectively enhanced tracking and guidance accuracy and guaranteed moderate level of computation load at the same time.