In order to solve the problem of GNSS signal loss caused by GNSS faults or external environment occlusion in INS/GNSS integrated navigation system,an integrated navigation algorithm based on Random Forest Regression (RFR) factor graph is proposed.Firstly,the Inertial Navigation System (INS) and Global Navigation Satellite System (GNSS) are modeled by using factor graph method,and the factor graph model of INS/GNSS integrated navigation is built.Secondly,the random forest theory is introduced to build the random forest,and the training is carried out when the GNSS signal is effective,and the GNSS signal output when the satellite navigation fails is simulated.Finally,a simulation experiment is designed,and the results show that the improved factor graph algorithm improves the navigation accuracy by about 10%~15% compared with the federated Kalman filter algorithm.Meanwhile,the proposed RFR factor graph algorithm can still maintain high accuracy in the case of GNSS signal loss.