Abstract: An observable degree analysis method based on the low dimensional SOM is presented for quantitatively analyzing the observability of the states in Kalman filter.The SOM is proved to possess the same observation effectiveness as the total observable matrix via the modeling of the PWCS,based on which the piece interval of the model is taken as the section partition unit of the SOM to reduce its dimension,and then the observable degree of system states is defined according to the singular values of the low dimensional SOM.To avoid the potential uncertainties of the datum value in the definition,the reliability of the value and its influencing factor are introduced in the definition,subsequently a new definition mode is proposed.Finally,an example of the passive Beidou/SINS deep integrated navigation system is used to illustrate the proposed method.