Redundant information would decrease as the reduction in GNSS satellites, resulting in degradation of RAIM performance. Aiming at this problem, a method of chi-square test of multi-samples based on the sliding and accumulation is proposed. With the help of INS, it is used to achieve fault monitoring by extrapolating GNSS observations at multiple time to certain observation at the same time for making their consistency test. The essence of the method is that the fault monitoring is achieved by ensuring redundancy of observations at the expense of real-time autonomous integrity monitoring. In the paper the algorithm is used for GPS to simulate and compare fault monitoring performance, and validated in applications during different civil flight phases. The simulation shows that the algorithm can improve the integrity of GPS and present a new solution to the application of RAIM in aviation.