To improve the positioning accuracy and stability of the positioning system in GNSS-denied environments, the paper proposes a UWB quality control method considering NLOS and system errors, and achieves its integrated positioning with GNSS / INS. Firstly, considering the stability and accuracy of the positioning system, a centralized Kalman filter is constructed using loose combination of GNSS / INS and a tight combination of UWB/ INS. Baesd on this, aiming at the NLOS error existing in UWB, a two-step NLOS error identification method based on sliding window and innovation vector of the filter is designed. Finally, a method based on filter estimation is used to compensate the system error in UWB in real time. The experiment results show that the proposed UWB quality control method can effectively reduce the impact of NLOS and system errors, and the horizontal positioning error of the GNSS / UWB/ INS integrated algorithm is within 5cm. With a reasonable UWB layout, this method can achieve high positioning accuracy without relying on an excessive number of base stations.